Twenty five formations of Greylag Geese Anser anser were photographed from immediately below to eliminate perspective distortion, and the lateral and longitudinal displacements of the birds relative to each other were measured. We scaled the photographs and used measurements of bill to tail and wing span made on 1 5 freshly shot birds to convert the lateral displacements to wing-tip spacings. The birds flew on average with an overlap of their wing tips (median = 17 cm) which corresponded very closely with the overlap expected (1 6 cm) from an aerodynamic model which predicted the position which maximized energy savings. However, the variation in positions was large, and only 17% of birds actually flew in the optimum range. The mean saving in induced power averaged across the distribution of positions was 26.5%, and the contribution to total flight costs was a reduction of 4.5-9%. This saving was greater than we found previously in the Pink-footed Goose Anser brachyrhynchus. There was no correlation between position in the skein and longitudinal displacement, as would be anticipated if the birds were equalizing the savings across the skein. This does not mean costs were not equalized because other mechanisms are possible. There was no correlation between depth and wing tip spacing which does not support the orientation/communication hypothesis. Body size and thus flight costs may be a factor influencing the function of formation flights. 2 8 0