264 H. LYNDHURST DUKE. 



On Tavu, therefore, the breeding accommodation for the fly is reduced to a mere 

 fraction of what existed formerly. The shrubs limiting the old breeding area on either 

 side are growing on sand ridges above the present water-level ; among' their roots a 

 few pupae were found. Even here, however, the places where the sand is sufficiently 

 •dry for the requirements of the pupae are rare. During a period of 55 minutes, 10 

 boys obtained 79 pupae working hard under my supervision and encouraged by 

 a bonus according to the numbers found. Coincident with and presumably as a result 

 of this temporary destruction of their breeding ground, the number of flies has been 

 greatly reduced, and during this spell of pupa-hunting only a very few were seen. 

 On a sunny day 3 boys working for 5 hours caught 140 flies, although offered every 

 inducement to obtain as many as possible in the time at their disposal. 



The most striking deficiency from the fly point of view at the present moment on 

 Tavu Island seems to be dry sand. 



Bulago Istond. 



On this island Fiske found pupae in fair, though not very large, numbers in 

 1915. Here the shore line is somewhat steeper than on the more favoured pupa 

 islands such as Tavu and Damba. As a result, the lake rise has produced relatively 

 less detrimental effect on the sandy breeding areas. From the strip of shore to the 

 North- West, where pupae were formerly most numerous, 5 boys working under 

 my supervision for 5 hours obtained 214 pupae. On another occasion 7 boys obtained 

 141 in 1 \ hours from same area. These were, as usual, in the sand among the roots 

 of the " nzibaziba " and " byansambwe " bushes, and were only found where the 

 soil, either sand or dusty humus, was dry to a depth of 2 inches below the surface. 

 The great majority of the pupae were found below the high- water mark, which, as will 

 be seen later, corresponds to the lake level of June 1917. Often the sand under the 

 bushes appeared at first suitable for pupae, but on investigation proved to be distinctly 

 moist immediately below the surface, and no pupae were found. Occasional pupae 

 or empty cases were found in the zone immediately above the 1917 high-water level, 

 either in sand or in dry dusty earth in the forest belt at the bases of the trees. As 

 Carpenter points out, ant-lion larvae are f ound in the same type of soil. 



The flies on this island, in my experience never very numerous, are perhaps less 

 reduced in numbers than on other neighbouring islands. In all 196 flies were caught 

 in Bulago the sex proportions being males 84 '70 per cent., females 15 - 30 per cent. 



Nsazi and Kimmi Islands. 



Carpenter (1) writes of Nsazi : "At only one place along its coast are flies at all 

 numerous." This is along the sandy shore of a bay on the S.W. side of the island, 

 where a fine stretch of dazzling white sand fringed with bushes is backed by a zone 

 of forest which runs up to the base of a steep cliff. Pupae were formerly found here 

 in small numbers. Between 22nd January and 10th February 1911, Fraser and I 

 carried out experiments with the wild fly of these two islands. The figures of the 

 catches made by 6 fly boys during this period, two of whom accompanied me in July 

 1918, were 5,765 from Nsazi and 8,444 from Kimmi. On Kimmi, where the fly 

 used to be exceedingly numerous, the boys were limited to 1,200 flies per diem, as the 

 available experimental monkeys on which they were fed could not stand a greater 



