The pattern of density of the microfilaria Chandlerella quiscali among host specimens of the gnat Culicoides crepuscularis was fitted to the negative binomial distribution. Variations in the density of microfilariae in peripheral circulation of the grackle Quiscalus quiscula versicolor were suggested as the reason for the clumped density of nematodes among the gnats. Analyses of variations in pattern of distribution have been applied to plant and animal populations to gain insight to some of the biological interactions that vary over time and space: (e.g., Kershaw, 1964; Greig-Smith, 1964; Pielou, 1969; and Crofton, 1971a, b). Three general categories of pattern may be defined as (1) regular, (2) random and (3) clumped. Each category can be tested for fit by observed sets of distribution data, and each infers certain kinds of ecological phenomena. Deviations from the random (Poisson) pattern imply an interdependent relationship among individuals or between organisms and some environmental (habitat) variable. For example, regular patterns may be seen in species that saturate habitats to the level where minimum, equal distances separate all individuals. A clumped or contagious distribution pattern can be seen where habitats are not uniform and individuals are concentrated in the most suitable patches, or where interactions among individuals may hold organisms together in groups although the habitat is uniform. Analyses of parasite populations for variations in pattern of distribution among and within host specimens provide a means of defining certain categories of parasite-parasite or parasite-host interaction, and of extending our ability to describe symbiotic relationships in general. This report describes the analysis of pattern of distribution of microfilariae (Chandlerella quiscali) among host specimens of the gnat (Culicoides crepuscularis). MATERIALS AND METHODS The data for this analysis came from the study by Robinson (1971) on the role of the ceratopogonid gnat Culicoides crepuscularis as a host for the filarial nematode Chandlerella quiscali that Received for publication 31 March 1972. parasitizes the purple grackle, Quiscalus quiscula versicolor. In that study, 2 kinds of traps and a variety of trapping times were used to expose and collect the Culicoides. However, the data for the present paper include only 143 gnats from the total collection, that were taken under a single set of standard conditions. The same infected bird was used for all exposures. The microfilarial density in its peripheral blood during the hours of trapping was not determined, but relative to infections studied in many other birds this one was light to moderate. The trapping time was 2100 to 2300 hr, a period of the day in which it was known that gnats were maximally active, and in which the nocturnally periodic microfilariae were swarming in the peripheral blood of the host. Traps were run on nights in June and July. A modified funnel trap was used, in which 12 birds were contained as attractants, but the gnats could only reach the infected bird to feed. Three of these traps were run in different ecological situations and heights, known from previous work to be well populated by the gnats. Exposed gnats were separated from the unengorged ones, and maintained on honey water in small cages until killed for dissection and examination. There was no detectible mortality among the gnats during the developmental period of the worms, and likewise no mortality among the nematodes. Unengorged gnats served as controls and none contained microfilariae. The large number of gnats, n = 143, pr vided a sample size sufficient for statistical treatment. The raw data were tabulated in three ways. First all of the 143 gnats sampled were treated as a set for calculation of a sample mean and variance. Second, only those 85 gnats containing nematodes were tested for fit to a truncated (the zero frequency class was not included) Poisson distribution. Finally, all 143 gnats were separated into frequency classes by the numbers of contained nematodes and statistical testing was done for fits to the Poisson and the negative binomial (clumped) distributions.