Numerous studies have examined sexual size dimorphism in birds and speculated upon the reasons for its existence. Whilst most studies have focused on individual species or groups of related species, a few have attempted to disentangle the various hypotheses that have been put forward to explain its occurrence. Typical of the latter studies is that by Jehl and Murray (1986), in which they argued that sexual size dimorphism was primarily the result of sexual selection (see also Bennett & Owens 2002). Although some studies have looked at patterns in sexual size dimorphism without calculating a figure to represent the difference (e.g. Amadon 1959), most have examined measurements of birds and used these to calculate such a figure. Traditionally in such studies, measurements used have included wing-length, culmen-length, tarsus-length and mass, although McGillivray (1989) took the sum of 18 skeletal measurements and used their male and female means to determine sexual size dimorphism. Wing-length has commonly been used to determine sexual size dimorphism, although lack of repeatability of measurements may render it less useful than skeletal measurements like tarsus-length as in studies of Dunlin Calidris alpina (Blomqvist et al. 1997) and Savannah Sparrows Passerculus sandwichensis (Rising & Somers 1989); however, Gosler et al. (1998) found wing-length measurements to be more repeatable than other metrics in a group of 27 passerines.