1. The filarial worm Brugia patei was brought to London from East Africa in the larval stage developing in Mansonia mosquitoes. Subsequent transmission to domestic cats was possible in London through a laboratory colony of Mansonia uniformis and the worm was very infective to this species of mosquito. The worm was much less infective to Aedes togoi and Anopheles gambiae.
2. Subsequent passage of the worm through Aedes togoi has increased its infectivity to this species of mosquito. The infection rate has risen from 43·6% in females of A. togoi fed on the cat original- ly infected by larvae from Mansonia mosquitoes, to 59·2% in females fed on a cat containing the first generation of worms to pass through A. togoi, to 79·1% in females fed on cats containing the second generation of worms to pass through A. togoi, to 86·7% in females fed on cats containing the third generation of worms to pass through A. togoi, to 89·8% in females fed on cats containing the fourth generation of worms to pass through A. togoi.
3. The distributions of infective stage larvae in female mosquitoes were negative binomial distributions. These were related to the distributions of microfilarial intake immediately after infection which were also negative binomial. From one generation of worms passed through the new mosquito host to the next the mean and the k value of the negative binomial of the infective stage larvae approached the same parameters for the distribution of microfilarial intake. By the fourth generation of worms passed through A. togoi the distributions of microfilarial intake and infective stage larvae developing in A. togoi were similar, suggesting close adaptation to the new mosquito host. A similar relation between these distributions was found originally in Mansonia. There were however considerable differences in these distributions between the two species of mosquito.
4. The experiment extended over a period of 6 years. The change in infectivity did not appear to be related to the microfilarial density, age of infection, or sex, in the mammalian host. Neither could it be related to the genetics of susceptibility in the mosquito, nor to mosquito culture techniques, nor to mosquito mortality.
5. It is concluded that the observed change in infectivity is best explained by a change in adaptation of the worm to the new intermediate host by selection within the mosquito of those microfilariae adapted to it. No major change in infectivity to the original intermediate host, Mansonia, was found during the experiment.
6. These findings are examined with reference to changes in adaptation to mosquito hosts in the history of Bancroftian filariasis.