In a field study, we show that a young song sparrow (i) selects his songs from three or four older birds who have neighboring territories, (ii) preferentially learns song types that these tutor neighbors share, and (ui) ultimately sets up his territory next to, or replaces, one of these tutor neighbors. The consequence of this song learning strategy is that the young bird's song repertoire represents the "logical intersection" of the song repertoires of his tutor neighbors. We argue that this repertoire is optimally designed for mimicry (sounding like your neighbors) and for communication between neighbors (song sparrows address or reply to a neighbor with a song they share with that neighbor).Song learning in passerine birds is a selective process in which the young bird retains in his final, adult repertoire only a fraction of the many song types to which he is exposed (1). Despite considerable theoretical interest in the design and function of song repertoires (2-6), however, little is known about the variables determining which of the "tutor" songs are selected for the song repertoire, other than that conspecific songs are preferred over heterospecific songs. To test the hypothesis that social variables are the key determinant of song selection, we studied song learning in a free-living population df song sparrows (Melospiza melodia).In this study, we tested a model of song learning derived from three sets of observations. First, in many (but not all) songbird species, birds share their song types with neighbors, with the resemblances in some cases being so close as to Suggest that one bird learned the song type from the other (7). Second, young male song sparrows, starting in the summer of their hatching year, "float" on the territories of several adjacent territorial males and eventually (usually by the following summer) try to set up a territory in this floater range (see refs. 8 and 9; unpublished observations). Third, in song sparrows (and many other songbirds) the early part of a bird's life (especially his second and third months) is critical in the formation of his song repertoire (10). Since we have never observed a male song sparrow to add of drop a song type between his first breeding season and subsequent years, we assume that a.young song sparrow's repertoire crystallizes sometime in his first year of life, possibly as early as his hatching summer. Putting these observations together, we hypothesized that during this floater period the young bird learns the song types of some or all of the territorial males in his floater range. To test this hypothesis, we attempted to trace the song tutors for a sample of young male sparrows from our study population.In this paper, we follow conventional usage and call the bird from whom the song was learned the tutor (the learner is the student). We do not mean to imply by these terms that the older bird actively teaches the younger bird (although he may) or that the younger bird is a passive learner (indeed our evidence suggests quite otherwise Wash-...