The Origins of Control: Air Pollution and the American State, 1910 -- 1938

Abstract

The United States government produces a wealth of data on air pollution. This dissertation asks how this is possible in a state known by political theorists for its decentralized administration of official knowledge. To answer this question, I explore the origins of the "environmental state," between 1910 and 1938. The official knowledge used to craft legislation such as the Clean Air Act was first developed in this period in an "associational" concert of statal organizations in the federal government, academia, professional associations, and industry. Other publics, including exposed communities and labor, were largely left out of the selection and recombination of older ways of thinking about the smoke, dust, and gas problems to address the new problem of air pollution. The result was that community struggles over air pollution problems were channeled into epistemic puzzles that could only be resolved by more and better monitoring systems. This, in turn, led to an increasing emphasis on air pollution monitoring, rather than control, and a surveillance infrastructure that was not matched with commensurate regulatory powers. I come to this conclusion by tracing the infrastructural and institutional preconditions for this pattern. The infrastructure has its origins in the British national air monitoring network first developed in 1912. I follow the development of this network, and compare its dynamics to those that prevailed in the United States. Variation in the relations between central governments organizing the networks and the municipal contexts in which the monitoring takes place result in a distinctive composition and development strategy in the United States. Where the British network emphasized longitudinal and geographic stability in its data production, the US network emphasized repeated technical experimentation in the expectation that practices would be taken up as their worth was demonstrated. This did not occur. Instead, repeated failures for a US network to consolidate finally broke in the rapid assembly of a national monitoring network after 1953. Rather than adopting the new methods for themselves, local leaders instead demanded inclusion in the (by now) well-practiced air pollution monitoring organizations directed by federal administrators. This pattern was made possible, I claim, by certain institutional preconditions. First, an ideology of associationism came to dominate executive branch organizations in the interwar period. This ideology, best represented by Herbert Hoover's technical writings on mine engineering, would become one side of a channel shaping environmental statecraft in the years to come. The other side of the channel was formed from the relations of knowledge production in the federal organization primarily responsible for air pollution problems between 1910 and 1938, the United States Bureau of Mines. Hoover's associationist ideas were projected through the contingencies of the federal bureaucracy over these years, in which warm relations with industry needed to be maintained in order to secure streams of talent and legitimacy for the organization. Together, these channels shaped a style of atmospheric governance that tended to view consciousness of problems as leading inexorably to their control (if not its equivalent). I move through two layers of archival materials to identify these patterns. The first layer is usually, but not always, digital. I sort and carefully read the massive online archive of technical documentation, bulletins, reports, articles, proceedings, and other bureaucratic paraphernalia to identify important actors and their relations. In key areas, it is necessary to move to a second layer of archive that is not usually digitized, though important exceptions exist. This second layer consists of the metadata for the first, including the correspondence, rough drafts, meeting minutes, data, memoranda, unpublished related work, and transcripts surrounding a given research project. In both cases, I combine eyeball readings with computational readings, in particular to provide distributional portraits of populations of texts in which key pieces of evidence are buried.