The design of subnational administrative units shapes the spatial organization of state bureaucracies and countries’ political topographies. Yet, we lack theoretical and empirical knowledge on how state territories are divided into subnational governance units. Financed by a Leverhulme Small Research Grant, this project argues that governments’ preference for decentralizing power determines the alignment of administrative borders with ethnic geographies in multi-ethnic states. Centralizing governments will draw borders that crack groups into diverse units, while decentralizing governments ethnically align borders, packing groups into homogeneous divisions. I will test this argument by studying unit-design in colonial Africa contrasting centralized, direct French rule with decentralized, indirect British rule with newly digitized geographic data on precolonial ethnic and colonial administrative geographies. I assess the conditional effect of ethnic settlement patterns on administrative borders using a spatial partition model which I have recently developed. With novel theory, data, and methods, the project contributes to our understanding of the origins of administrative geographies.