Beginning in 1920, bird banding in the United States was coordinated by an office within the U.S. Biological Survey that recruited volunteers, issued permits, distributed bands and reporting forms, and collected and organized the data that resulted. In the 1920s and 1930s, data from thousands of volunteers banding millions of birds helped ornithologists map migratory flyaways and census bird populations on a continental scale. This essay argues that the success of the bird-banding program depended on a fragile balance between the centripetal effects of national coordination and the centrifugal effects of volunteer enthusiasm. For various reasons, efforts to maintain this balance were largely abandoned by the Bird-Banding Office from the late 1930s onward. Nevertheless, the first two decades of the national bird-banding effort provide an example of how a "citizen-science" project that generates "Big Data" can produce significant scientific results without subordinating the enthusiasms of volunteers to the data-collecting needs of professional scientists.