Partial statistics: selectives and illegibilities in 19th-century censuses

Abstract

This article examines how strategic disappearances and selective visibilities became embedded in the management of census data in Brazil between the late nineteenth and early twentieth centuries. By tracing marginal annotations in census forms, ambiguous categories, budgetary constraints, and fragile institutions, the text investigates the processes through which illegibilities were produced in the country’s early national censuses, from 1872 to 1920. The analysis is grounded in an ethnography of documents and archival materials related to the imperial censuses, based on research conducted at the Brazilian National Archives, the Brazilian Institute of Geography and Statistics, and the Federal Senate. The findings of the study contribute to scholarly debates that understand imprecision, inefficiency, ambiguity, and error not merely as flaws but as forms of data governance embedded in the very design of statistical models. The article further suggests that such forms of selective visibility and illegibility operate as mechanisms for arbitrating access to rights and resources, shaping the governance of populations and territories. By shifting attention to the zones of illegibility and the material conditions that produce them, the article invites reflection on how the state sees—and fails to see—its population through the use of often inaccurate or outdated statistical data.