Some understanding of the vertical disposition of flying insects is essential to interpret their movement from observations made near the ground. Ground-level observations of the migration of moths are often made with light traps because Lepidoptera are comparatively sparse and, to catch enough for analysis, the aerial population must be artificially congregated at the sampling site. Light traps do this admirably and, provided the congregating process is measured or at least understood, the resulting samples are useful for many purposes. Unfortunately moonlight creates an additional variable in light-trap sampling, apparently by affecting the congregation of moths at light in ways not well understood; hence there is usually a pronounced lunar periodicity in numbers caught. Whilst it seems certain that this periodicity is in part an artifact of trapping, the possibility remains that moonlight may affect, not only efficiency of sampling, but also the proportion of the flight-worthy individuals of a population that fly, and their vertical disposition. Since Williams (1936) there have been several attempts to distinguish artifacts from direct effects of moonlight on flight, but with moths it is difficult to find an independent standard, unaffected by light, against which to compare these effects. Suction traps offer a method of non-selective sampling but, unless the traps are impractically large, the sample of moths is numerically small. Also, because light traps selectively discriminate between species, those species most effectively congregated by light may be hardly represented in suction trap catches (Taylor & Carter 1961). Hence direct standardization of light-trap catches of moths by suction or other similar traps is difficult. Standardization of light-trap catches of insects smaller and commoner than Lepidoptera has been more successful because both kinds of traps yield numerically large samples. In particular, excellent comparisons of catches of mosquitoes from light traps with those from suction traps, which give a point sample, and from truck traps, which sample a transect (Provost 1959; Bidlingmayer 1964, 1966, 1967), have shown that the decline at full moon of light-trap catches near the ground is largely an artifact. Williams, Singh & el Ziady (1956) adopted a different approach. Comparing suction trap samples at 5 ft (1 -52 m), in garden sites exposed to moonlight, with those in a woodland where the effect of moonlight was minimized by the canopy, and with those at 30 ft (9414 m) in the open, they sought differences in the vertical and horizontal disposition of insects at new and full moon. Although there was evidence of a lunar cycle of numbers at 5 ft in the open during the autumn of 1 year (Williams & Singh 1951), 3 succeeding years'