In 1930, a time when human geography was not considered to be among the emerging analytic social sciences and when urban geography barely existed as an identifiable subfield, C. Warren Thornthwaite, later a gargantuan figure in American climatology, was awarded a doctorate under Carl O. Sauer at the University of California, Berkeley. In his dissertation, ‘Louisville, Kentucky: a study in urban geography’, Thornthwaite analysed land-use zonation, commercial strip development, daytime–night-time populations, community formation, elevation and land values, functional regions and cultural areas, spatial competition and optimal location, buying power and residential distribution. Thornthwaite's urban geography was centered unmistakably in the spatial tradition, and his innovative research – based on aerial photographs, intensive field observations, city directories and detailed mapping – treated concepts that were to become central to urban geography as it developed in the late 1950s and the 1960s. Urban geography, however, evolved independently of Thornthwaite's contributions. Thornthwaite directed no graduate students, and his unpublished dissertation was not cited until a perfunctory reference in 1954. A second citation, not appearing until almost 30 years later, described it as ‘a major synthesis’. Thornthwaite's dissertation was not only an anomaly in the career of its author but also an anachronism in the history of urban geography.