1. Work on the mosquitos frequenting native huts was carried out at Kisumu, Nyanza Province, Kenya, during 1940–41.

2. During preliminary night and day catches in native huts the catching methods were standardised. Subsequently four identical huts were built side by side for comparative work. They were designed specially for easy catching and had hessian ceilings. Apart from catches made for special purposes, 6 day catches and 1 night catch were made every week in one of the huts for an entire year. This hut was regarded as a control. Records of climate were kept in this control hut for a year, the readings being compared with readings taken in a Stevenson screen outside.

3. It was found that the temperature range in the hut was only about half that in the screen—in other words, a native hut is a very well insulated structure. In hot, dry weather there were considerable differences between indoor and outdoor temperatures but in cool wet weather the differences were very small. Thermograph records showed that the form of the daily temperature curve in wet weather was quite different from that in dry weather. Temperature in the hut very rarely fell below the level at which development of the malaria parasite in the mosquito is arrested. In the screen it very frequently did so in the hours before dawn.

4. Hygrograph records of relative humidity in the hut and screen showed much the same characteristics as the temperature readings. Again the range in the hut was much smaller than that in the screen and differences between indoor and screen humidities were great in dry weather and small in wet weather. The form of the daily curve in dry weather differed markedly from the wet weather curve. Saturation deficit at night was always greater in the hut than outside.

5. At night, the climate of a hut containing 10 natives was compared with that of the control hut, containing 1 native. It was found that the crowded hut was slightly warmer than the other and that its relative humidity was somewhat lower than that of the control hut.

6. High wind at night was unusual. When it occurred the entry of mosquitos to huts was stopped for the time being.

7. A list of all species encountered is given, with locality records and sites of capture.

8. From the results of night catches the entry curves of various species were worked out. It was found that in the case of A. gambiae females there is a high population of resting mosquitos at 7 p.m., a period of low entry from 8 to 10 p.m., a period of high entry from 11 p.m. to 5 a.m., and a sharp entry peak at 6 a.m., due to the arrival of females seeking shelter for the day. The same remarks apply to A. funestus females, but the time distribution of the entry periods is different, namely :—7 p.m., a high population of resting mosquitos ; 8 p.m.–2 a.m., a long period of low entry ; 3–5 a.m., a short period of high entry ; 6 a.m., a large entry of females seeking shelter. In the case of both species, most of the females taken were fed, and there is close correlation between the curves for fed and total females. It is thought that females remain in the hut after feeding. Mature females with Group V ovaries enter mainly at dawn. Males of both species are scarce during the night but enter in large numbers at 6 a.m. to rest during the day. A. coustani and A. pharoensis are scarce in the evening and at dawn. A. coustani shows a well defined entry peak from 11 p.m. till midnight and A. pharoensis from midnight till 2 a.m. T. africanus and T. uniformis act rather similarly. T. uniformis shows a very distinct entry peak at 9 p.m. and T. africanus from 9 to 11 p.m. There is close correlation in this case between unfed and total females. It is thought that these species leave the hut soon after biting and that their biting activity is at a maximum between 9 and 11 p.m. Thus the entry curves of the important species show a succession of entry peaks through the night—T. uniformis, 9 p.m. ; T. africanus, 9–11 p.m. ; A. coustani, 11–12 p.m. ; A. pharoensis, 12 p.m.–2 a.m. ; A. gambiae, 11 p.m.–6 a.m. ; A. funestus, 3–6 a.m.

9. Night catches, comparing dark and dimly lighted tents, showed that A. coustani entered tents in large numbers and preferred the lighted tent, as did T. africanus and T. uniformis. C. nebulosus preferred the dark tent. Catches made in the open showed that A. gambiae and A. funestus prefer to bite in huts, T. africanus and T. uniformis in the open.

10. Comparison of night and day catches showed that the mosquitos taken may be divided into 4 groups :—

1. True house-haunting species, A. gambiae and A. funestus. Females are abundant both by night and by day, males by day only. These species enter both for food and for shelter. Day catches, which included both males and females, give the best estimate of prevalence.

2. Species which enter mainly to feed, T. africanus and T. uniformis ; perhaps A. coustani and A. pharoensis. Females are common by night but are never numerous by day. Males are always scarce. Night catches are essential for an understanding of this group.

3. Species which enter human habitations mainly for shelter, C. antennatus. Females are prevalent by day only.

4. Species which enter casually and are scarce both by night and by day.

11. Analysis of day catches showed that the great majority of A. gambiae and A. funestus females had Grade 1 wings. Sometimes the wings were still undamaged when the ovary was maturing for the second time. Ovary dissections of these species showed that the largest numbers belonged to Group I and that the percentage in this group became higher at times of increase. Most females of these species and of T. africanus and T. uniformis contained blood. Of the unfed A. gambiae and A. funestus, the great majority had Group I ovaries, i.e., they were young females which had never yet fed.

12. Analysis of an A. gambiae increase (December 1940–January 1941) by ovary dissection of day catches showed that the whole increase was dependent on a constant supply of young females with Group I ovaries. As soon as this supply diminished the increase came to an end. It seems, therefore, that the life of the female was short at that season. The numbers of males and females taken were closely correlated throughout.

13. In day catches the great majority of Anophelines were taken on the ceiling. Taeniorhynchus spp. were commonest on the walls. Very few mosquitos were found on the floor. With regard to mosquitos taken on the walls, it was found that all species preferred the lower and darker half. Catches made on the outside of the walls consisted mainly of male Anophelines.

14. In comparative series of catches made to determine the best hour for day catching, no significant difference was found between catches made at all hours from 6 to 11 a.m.

15. Preliminary tests showed that there was probably a quantitative relationship between the numbers of human inhabitants of a hut and the numbers of mosquitos taken. This was confirmed by night and day catches made simultaneously in huts with 1, 5, 10 and 15 inhabitants respectively. It was found that there was a very close direct relationship between the size of the human population and the numbers of females of biting species taken. Non-biting species did not discriminate between the huts.

16. Comparison was made between huts containing 5 washed and 5 unwashed natives respectively. All the important biting species preferred the hut with the unwashed natives.

17. In night catches, A. gambiae and A. funestus females preferred a hut containing dirty clothing only to a completely empty hut. Females of T. africanus and T. uniformis showed no such preference.

18. Catches in empty huts showed a relative but not an absolute increase in Anopheline males.

19. It was found that a hut subdivided by a hessian curtain could be used for preference experiments with varying human population.

20. In a comparison between a hut containing a native and one containing a calf, it was found that A. gambiae and A. funestus were more numerous in the hut with the native both by day and by night. A. funestus showed a more marked preference for the native than did A. gambiae. A. coustani, A. pharoensis and C. antennatus showed a distinct preference for the calf. In the hut with the native, Anophelines formed 94 per cent. of the total catch, but in the calf hut Culicines formed 95 per cent. of the total.

21. It is shown that at Kisumu the seasonal incidence of A. gambiae is dependent on local rainfall. The relationship is qualitative rather than quantitative. A threshold figure of 5 inches of rain per month is necessary before an increase can take place. When an increase does occur, the population in the first month of the increase is proportional not to the amount of rain which has fallen but to the lower initial population of the preceding month. If the increase is maintained for several months, the growth of the population takes place in geometrical ratio.

22. At Kisumu the seasonal incidence of A. funestus is not connected with local rainfall. Increases take place when the level of Victoria Nyanza rises. The rise in lake level is dependent on the rainfall of a large part of East Africa and is independent of local rainfall.

23. The seasonal incidence of other species is briefly discussed.

24. The parasites noted from mosquitos were mites, which were most prevalent on A. coustani and A. gambiae, and Coelomomyces africanus which was commonest in A. funestus. Coelomomyces is a fungous parasite which causes destruction of the ovaries. It is now recorded from Kenya for the first time. Unsuccessful attempts were made to infect larvae with this fungus, but naturally infected larvae were found.

25. Unsuccessful attempts were made to render adult mosquitos individually recognisable by staining the larvae with various dyes. Attempts to make the blood-meal recognisable by injecting identifiable substances into the blood-stream of the human bait were also failures.