Ten types of vocalizations of Purple Martins (Progne subis) from Texas and Arizona were described and compared. Solitary-nesting martins in Arizona did not possess greater vocal repertoire sizes than martins nesting colonially in Texas, which suggests that vocal repertoires are not evolutionarily plastic and have not changed with recent shifts toward coloniality in Texas birds. Some vocalizations differed markedly in structure between these populations, documenting for the first time, geographical vocal variation in the Hirundinidae. A re-examination of subspecific affinities of Purple Martins in Arizona seems warranted on the basis of voice; birds in mountain forests have vocalizations similar to those of birds in deserts, and both differ from the eastern nominate race. Purple Martins possess a greater vocal repertoire than do more colonial swallows, probably because historically they have been largely solitary and lacked acoustical constraints on the evolution of vocal repertoires. The Purple Martin (Progne subis) is a widespread North American swallow, nesting in backyard birdhouses in the eastern parts of its range and in remote high mountain forests and deserts in the West. Its breeding biology and behavior have received considerable research attention (Allen and Nice 1952; Johnston and Hardy 1962; Finlay 1971; Niles 1972; Brown 1978a, b, 1979, 1980; Brown and Bitterbaum 1980), but, aside from brief comments by Johnston and Hardy (1962), its vocalizations have not been described. This study was undertaken with three objectives. First, I sought to test the hypothesis (Brown 1983, in press) that solitary swallows have larger vocal repertoires than colonial species. Purple Martins, at least before the recent advent of birdhouses, were largely solitary, nesting primarily in abandoned woodpecker holes. Following Brown (1983, in press), one might expect larger vocal repertoires in martins than in more colonial species such as the Cliff (Hirundo pyrrhonota) or Bank (Riparia riparia) swallows. This study examined repertoire size in martins. Second, I examined whether either call structure or repertoire size differ among populations of Purple Martins inhabiting different geographical regions and exposed to different selective pressures. Martins nesting in suburban backyards in Texas, often in colonies of up to 50 pairs, were compared with birds nesting solitarily in mountain forests of Arizona. If martins in Arizona have larger vocal repertoires than those in Texas, this could mean that vocal repertoire size is evolutionarily plastic, possibly having changed with the recent shift toward coloniality in Texas birdhouse populations. Differing social structures could be very important in the evolution of geographical variation in avian vocalizations. Lastly, I wished to catalog and describe vocalizations for a genus about whose vocal communications nothing had been reported. STUDY SITES, METHODS, AND TERMINOLOGY I conducted this study in Sherman, Grayson County, north central Texas, and in the Chiricahua Mountains, Cochise County, southeastern Arizona. In Texas, most work was done from 1970 to 1979 at one colony (elevation 238 m) consisting of up to seven birdhouses (see Brown 1979, 1980 for details on study areas and how individual martins were recognized), with all actual recording done 5-22 April 1980 and 11-22 July 1980. In Arizona, martins were studied in 1980 at nesting sites in woodpecker holes of dead snags at Rustler Park (elevation 2,545 m) and near Herb Martyr Lake (elevation 1,924 m), both in the Chiricahuas, with all actual recording done 11-24 June. Field recordings were made with Uher 4000 Report L and 4000 Report 1C tape recorders and Uher M517 and Electrovoice Soundspot microphones, the former mounted in a 60-cm parabolic reflector. Tape speeds were 19 and 9.5 cps. Sonograms were made on a Kay Elemetrics Sona-Graph Model 6061-B using wide band pass setting and linear scale. Following Thorpe (1961) and Brown (in press), I used the following bioacoustical definitions: figure, a continuous tracing on a sonogram; syllable, any single figure or any two figures lasting 50 msec or less; song, a series of sounds of more than one type, uttered in succession and forming a recognizable sequence or pattern in time; subsong, an irregular and ill-defined series of syllables of lower in-