Abstract: Social interactions between individual horses were observed in two herds each comprising a stallion and a number of mares. In one herd, the animals were observed whilst grazing and resting; in the other, nearest neighbours were recorded when the animals were grazing, and social interactions were noted when the animals were feeding on hay.
In both herds, the horses showed marked preferences for the company of specific individuals when they were grazing. In one herd, the associations were mainly between individuals that had been associated prior to being put in the herd. In the other herd, this was not the case. A new statistic was produced for testing for specific company preference. In both herds, the stallion was dominant over all mares and never received any aggression.
The complete social hierarchy could not be determined for the herd which was observed only when grazing because social contact was restricted to that within groups or pairs that associated together. In the herd to which hay was fed, a non-linear hierarchy existed. Statistics were produced to quantify both the general level of dominance of a horse and its specific dominance or subordination to every other horse. It is suggested that these statistics, and one for quantifying the general aggressiveness of a horse, could be widely used.
A principal component analysis allowed the horses to be characterised socially according to aggressiveness, their attitude to other horses and their attractiveness to other horses.