FMGP sites represent a particularly pressing case for environmental researchers, sociologists, and communities alike. Despite the impact of the MFG industry on daily life, the built environment, and the industrial economy for nearly a century, contemporary social and environmental scientists have yet to address the industry as a systemic source of persistent urban contamination. Moreover, even as individual FMGP sites have dogged municipalities and agencies responsible for site remediation, researchers and remediation professionals are ill-equipped to track possible FMGP-related toxicity. The remains of the MFG industry stretch across every major and minor urban center in the United States, yet the discovery of FMGP sites is largely accidental (Hatheway, 2012). These challenges are mirrored in related research into legacy urban hazards more broadly. Three effects compound to produce the unfortunate result that the study of legacy urban contaminants has, until recently, proceeded with halting progress: Historically insufficient regulatory oversight, decades of redevelopment and urban change, and contemporary regulatory practices that focus on large and visible remediation efforts at the expense of smaller, pedestrian sites (Frickel and Elliott, 2018). The development of new methods to locate and assess possible sources of legacy contamination is thus a necessary first step to introducing a class of site and contaminant to spatial and sociological analysis.

For it to be both scalable and accurate, an effective site identification method must satisfy requirements of both scope and specificity. By scope, we mean that data should be available on a consistently wide range of geographies and time periods, and must be in a form that lends itself to rapid identification using easily-reproducible methods. By specificity, we mean that each site must be precisely located in space and time, and must be accurately identified. A district holding station, for instance, presents a different level of hazard from a central production complex, and must be identifiable as such. Due to limitations in available data, prior attempts at identification have largely satisfied either the “scope” or “specificity” requirement while neglecting the other. This division is facilitated by the major types of historical FMGP data that are readily available.

Prior attempts to locate FMGP sites on a broad scale have primarily relied on national gas industry directories, principally Brown’s Directory of American Gas Companies, published annually starting in 1887. Industry directories list the names of manufactured gas production companies, chief company officers, manufacturing statistics, and gas service areas, and provide a useful measure of the size and scale of gas production in any given city. But Brown’s and other directories do not identify street addresses for production or distribution sites. To precisely geolocate former gas infrastructure, industrial historians and environmental remediation specialists generally rely on historic building-level atlases, most prominently those produced by the Sanborn Map Company and other fire insurance mappers. Sanborn maps were produced for nearly all US cities with a population greater than approximately 8,000 (Hatheway, 2012). Sanborn maps were published as large, multi-page atlases, and include such information as building perimeter outlines and details on site use and ownership. Their high degree of street-level detail allow for the precise geolocation of gas production, storage, and distribution sites. For the same reason, however, it is impractical to rely on Sanborn maps to locate infrastructure at any appreciable geographic scope. To illustrate this, we briefly present two research efforts that typify approaches that focus on scope or specificity, but fail to precisely identify site en masse.

Our first study is the 1985 EPA effort to identify FMGP sites based on industry directory listings (Radian Corporation, 1985). Focusing on 10-year increments of Brown’s Directory publications, the EPA study estimated the number of large FMGP producers in the US at about 1,500. It was the first and only attempt at a national FMGP census, but - like other approaches that rely on industry directories alone - the EPA attempt was later shown to have greatly underestimated the number of possible FMGP sites nationally (Hatheway, 2012) for two reasons. Gas industry directories list only the cities in which major MFG producers were registered; second, directories do not indicate the number or location of gas industry production, storage, or distribution sites. Directory listings therefore drastically underestimate the scale of MFG’s spatial legacy, and do not provide sufficient information to indicate the precise street-level location of manufactured gas infrastructure. Using greatly expanded criteria, Allen W. Hatheway’s research estimates the total number of FMGP sites at between 33,010 and 50,308 nationwide, many times greater than the number of sites identified by the EPA (2012).

Our second example is a 1986 project that sought to locate the FMGP sites identified in the 1985 EPA effort (Ecology & Environment Inc, 1986). This second study was conducted only for EPA Region X (Alaska, Idaho, Oregon, and Washington), representing just 35 of the 1,500 sites identified the year before. Researchers identified site locations by pairing Brown’s Directory listings with Sanborn map data, and recommended sites for possible remediation based on information on current use. Study authors were aware that relying on Brown’s Directory information meant that they were likely missing district stations and other outlying utility sites. Their use of Sanborn map data nonetheless represented a clear extension from the broad-based approach of the 1985 EPA study, and study authors successfully located several dozen FMGP sites, several of which were shown to present possible hazards to human health. According to study documentation, however, locating map information for the 35 sites was budgeted at a full 86 hours - nearly two and a half hours per site. Researchers were also prepared with a full list of target sites prior to conducting street-level map searches; in contrast, the task of locating an unknown number of production, storage, and distribution sites by definition does not begin with a known, fixed universe of site names.

Both EPA efforts were shaped by the limitations built into their input data. Directory listings require little sifting to identify a broad range of possible FMGP sites within a given city or state, without providing information on where those sites may be located; Sanborn maps require significant investment to precisely locate specific sites, and benefit from having prior information on site whereabouts.

Source type Data Scope Pub. frequency Specificity
Industry directories Centralized National Annual Company name only
Sanborn maps Fragmented National ~10 year increments Multiple site locations