Whether animal populations are regulated by social behaviour at levels limited by their food supplies (Lack 1954) or safely below the threat of starvation (WynneEdwards 1962) is still in dispute. Prudent restraint (which Wynne-Edwards refers to as self-regulation or self-limitation) involves a degree of altruism that he thought unlikely to evolve under individual selection, so he invoked group selection (Wynne-Edwards 1962, 1986). The subsequent heated debate was largely defused by the recognition of kin selection and the everwidening concept of inclusive fitness (e.g. Mayr 1991, Bell 1997), and by the demonstration that at least some forms of altruism can arise under individual selection. However, the basic problem remains: does social behaviour regulate populations at, or prudently below, their food limits? This problem needs to be examined independently of its mechanism of evolution. A comparison of the population densities attained by species and their domesticated forms, whose social behaviour has been disrupted by long-term selection for tolerance to crowding given unrestricted food, could provide a measure of this effect of social behaviour. The remarkably different densities attained by European rabbits (Oryctolagus cuniculus (L.)) and brown hares (Lepus europaeus Pallas) are instructive. Rabbits often exceed densities of 100/ha, and can reach 200/ha on productive land and fertile islands (Thompson and King 1994). They have also been recorded to overeat their food supply and starve to death in enclosures (Gibb et al. 1978), on islands (Flux and Fullagar 1992), and on the mainland (Rolls 1969). Hares use the same grassland habitats as rabbits, and 78% of their diets overlap (Homolka 1987); yet hare populations seldom exceed 2/ha and the highest density ever reported, on a predator-free island with very favourable habitat, was 3.4/ha (Abildgard et al. 1972). As hares are only twice the weight of rabbits, they are obviously far below any significant food limit. This I demonstrated to my own satisfaction by keeping three pet hares in cages and feeding them for a year on vegetation collected from a 0.2-ha field, an effective density five times higher than any population has ever reached. Because rabbits and hares are important game animals, they have been introduced onto hundreds of islands, and frequently onto the same island in turn; so these highest density figures are based on a large data set (Flux and Fullagar 1992). The mean and range of the highest biomass recorded for each species of lagomorph listed by Chapman and Flux (1990) are as follows: for 12 species of Ochotona, 8.75 (0.7-38.0) kg/ha; for 14 species of Lepus, 8.96 (0.08-40.3) kg/ha; and for six species of rabbits excluding Oryctolagus, 6.28 (0.3-18.0) kg/ha. Clearly it is the European rabbit, with peak biomass reaching 330 kg/ha, which is exceptional rather than the brown hare at 13.6 kg/ha. The anomaly can be explained by Fitter's (1959) suggestion that the wild European rabbit was really a form, domesticated by monks for over 600 years, which is supported by recent DNA analyses (Thompson and King 1994). These rabbits were liberated on more than 800 islands (Flux and Fullagar 1992) and are the stock present in Australia, New Zealand, and South America, where their catastrophic effects on agriculture are well documented (Thompson and King 1994). A domestic animal is defined by Collins English Dictionary (1992) as one bred or kept by man as a pet or for purposes such as the supply of food; and feral as existing in a state, esp. after being domestic. A general review is outside the scope of this note, but ungulates seem to show the same pattern: goats on Macaulay Island, New Zealand, reached a biomass of 300 kg/ha (King 1990), while the biomass of all large herbivores combined in 24 African wildlife areas averaged 64.6 (4.1-199) kg/ha (Coe et al. 1976). The classic plague irruptions of species, such as lemmings,