Three aerial applications of the n-butyl ester of 2,4,5-trichlorophenoxyacetic m acid (2,4,5-T) in diesel oil were made at intervals of six weeks to an isolated woodland community on a peninsula in Lake Victoria, Kenya. The woodland contained two species of tsetse fly, Glossina pallidipes Aust. and G. palpalis (R.-D.) (subsp. fuscipes Newst.). The object of the treatment was to defoliate the woodland in order to ascertain whether the reduction in shade would control the tsetse fly. It was also desired to obtain further information on the effect of the spray applications on the woody plant species.

The first application, at an estimated actual dosage of 2·2 lb. 2,4,5·T per acre, resulted in a rapid defoliation of 13 out of the 15 observed leafy species, eight of, which subsequently died during the following nine months. Two succulent species of Euphorbia also died during this period. Of the remaining two species, one was fully defoliated only after the third application but recovered slowly, while the other, Lecaniodiscus vaughaniae, was but slightly affected at any time.

No significant reduction in the numbers of tsetse fly was observed during the nine-month period following the first spraying. The following factors probably account for this lack of effect: the varying rate of defoliation of the different plant species, the resistance to the spray of one locally abundant species, the persistent shade cast by the stems of the thick vegetation after defoliation and the relatively humid climate during the period of maximum defoliation. These factors allowed an environment favourable for the tsetse fly to exist in one part or another of the peninsula in spite of the severe effects on the vegetation.

Nine months after the first spraying, between a third and a half of the vegetation was destroyed during the dry season by an accidental fire. The effect on G. pallidipes was almost immediate, and two months later the population had been reduced to a fairly low level, which was maintained for the last 16 months of the experiment. The fire hardly affected the environment of G. palpalis and the main effect of defoliation was to cause this species to concentrate near the shady lake shore, where the dominant tree was Lecaniodiscus vaughaniae.

There is little doubt that the destruction of the vegetation by fire was made possible by the effects of the spray applications, as the hitherto evergreen thicket would normally have been unburnable. While this suggests a possible new technique for bush clearance, further trials have indicated several factors which may make difficult a satisfactory burn following defoliation.

Details are given of the assessment of the spray droplet performance both in calibration trials and during the applications to the peninsula.