Concise, critical reviews of books, exhibitions, and projects in all areas and periods of art history and visual studies
April 23, 2018
Yale University and National Endowment for the Humanities Photogrammar: Yale University and National Endowment for the Humanities, 2017.
Screenshot of Photogrammar homepage, April 24, 2018

Between 1935 and 1944 the US Farm Security Administration and the Office of War Information (FSA-OWI) commissioned a collection of 170,000 photographs. Ostensibly a public relations project to promote Franklin Delano Roosevelt’s resettlement programs for poor farmers during the Great Depression, they are a record of rural life and economic anxieties that were mediated by an intervention of industrialized public services. Now these images, along with some later additions, are a digitized collection of photographs and metadata that have been archived by the Library of Congress (LoC). This collection is also the primary object of inquiry on a digital platform called Photogrammar, a browser-based suite of tools that can be used to investigate the large photographic data set.

Photogrammar’s interface includes a number of customizable search features that are augmented with visualization tools to give the user agency in engaging with the photographs. Produced by a team of scholars at Yale University with support from the National Endowment for the Humanities, Photogrammar demonstrates the public impact that digital scholarship can facilitate by creating knowledge networks among historians, artists, archivists, and technologists. Team members include primary investigator Laura Wexler, a specialist in gender studies and the history of photography, and codirectors Lauren Tilton, an Americanist specializing in digital public humanities, and Taylor Arnold, a statistician.

Photogrammar is essentially an interactive catalogue raisonné, classifying all of the images in a particular archive so that one can study in them clusters to make comparisons. Art history research has always utilized genres of classification, but traditionally, the taxonomies were analog and static. With interactivity built into the interface, the viewer has agency to become more critical of the roles of images in our increasingly visual culture and of how our engagement with them contributes to both digital and visual literacy. It invites an understanding of images and cataloging techniques as knowledge generators in their own right.

Tools like Photogrammar allow researchers to make arguments with publicly accessible collections in order to demonstrate, with both case studies and large data sets, the impact of programs like the FSA-OWI that are documented in the LoC. For example, Sheldon Dick’s 1937 image of autoworkers’ houses in Flint, Michigan, [link] shows several sparsely planted, snowy yards (only one with a car), and laundry blowing on a clothesline on an almost certainly freezing day. A caption notes that rent was twenty-five or thirty dollars a month for the houses, and “water is drawn from a commercial well.” One can imagine the icy water that had to be transported in the snow to scrub the crisp laundry. That the photograph—an elegantly composed and well-documented window into the existence of unseen inhabitants—is now accessible to a wider public thickens our collective historical knowledge of a town that is currently facing a water crisis amid economic hardship.

The Photogrammar web interface feels similar to other data visualization tools with multiple functions, such as Tableau, a text-analysis tool geared toward businesses, or Palladio, an open-source tool for visualizing complex humanities data. To use either of those, the user uploads her own data set and chooses from several kinds of visualizations (such as a map, network, or gallery) in order to analyze the data. But in the case of Photogrammar, the content has already been curated and the user simply chooses which functionality to implement when examining the photographic data set. This tackles a fundamental issue in open-access research, which is that seemingly infinite amounts of openly available data are often difficult to discover.

One of the strengths of Photogrammar is its lack of disciplinarity. Its purpose is to expose and recontextualize an existing photographic archive, one with images that are works of art as well as documentary cultural artifacts. This is generally an unnecessary distinction, to be sure, but here it highlights the variety of ways scholars investigate images as sources: the digitized photographs can be interpreted as art, artifacts, cultural heritage, government propaganda, or documentation of daily life. The Labs section allows users to explore the images as any or all of these, which is crucial because the project engenders conversations about the many ways images contribute to knowledge. It offers both participation in and resistance to the concept of an impersonal, unreachable sea of big data. By treating images as data, Photogrammar makes possible the distant reading of a collection of photographs. Although the data set is large, the homogeneity of it allows the user to explore metadata and described content, which are both texts. The impending ColorSpace tool promises some non-textual analysis by way of color, hue, and saturation, but it remains to be seen whether that will be an informative avenue of inquiry since many of the images are black and white.

The interactivity of Photogrammar means that you can also investigate a “category” to see how many photographers were interested in a topic. The Metadata Explorer is currently available only for California, but you can use it to see that Dorothea Lange, Arthur Rothstein, and Russell Lee documented “religion” in that state. You can also use the county-based US map to discern at a glance that East Coast photographer Esther Bubley rarely crossed the Mississippi, but Lee worked in almost every region of the country.

One author used the tool for a historical investigation into the shooting practices of photographers on location by digitally curating disparate photos back into cohesive rolls of film, showing the photographer’s method or mind-set throughout a given project. In a blog entry titled “Strips,” the author describes matching identical call numbers in images’ LoC metadata to discern not only which images were on the same film rolls, but the order of shots taken in a given roll. They determine, for instance, that John Vachon’s attention shifted from novelties of daily life in Kenosha, Wisconsin, to urban poverty as his trip to Chicago unfolded.

Photogrammar is a few years old, so it is worth acknowledging why a review would be relevant now. As a project evolves or degrades, it is important to measure not only technical capabilities but ongoing impact—ways in which it serves as a springboard for discovery, research, or other projects. With the exception of some link rot in the Team section, Photogrammar’s functionality is still intact. Maps run on Leaflet (an open-source JavaScript library) and a database in Carto (a commercial web-mapping application) using open-source code to link their functionality. Photogrammar would be an interesting case study for The Socio-Technical Sustainability Roadmap (created in the University of Pittsburgh’s Visual Media Workshop), a compendium of digital-preservation guidance toward the long-term stewardship of digital projects. Photogrammar sparked curiosity and left me wanting to know more about how the tool was built, for how long it is funded (and whether “coming soon” sections such as ColorSpace or additional states on the Metadata Dashboard map are aspirational or in preparation), as well as the author of each blog post.

It is important to acknowledge the aspirational and ongoing nature of digital work­—while an “unfinished” project may be uncomfortable for some users, the potential to grow and expand is one sign of a productive and generative endeavor. Likewise, a project does not have to exist or degrade in a vacuum. Because Photogrammar is mostly open-source, this means that the underlying code can be copied and adapted by anyone who wants to make a new version, either in its current form or as a tool to visualize other photographic collections. I like the open-ended sense of discovery that comes from using Photogrammar, but if it were to receive an additional round of development (through funding or the open-source community), it would be helpful to include evidence of the tool’s impact: links to projects, publications, or use cases that emerge. More how-to posts for researchers like the “Strips” investigation would be ideal, as would suggestions for finding additional information, such as where to download the entire data set from the LoC or whether you can get higher-resolution photos through this interface. A reading list or bibliography about the photo collection, or even keyword suggestions, to channel visitors toward new avenues of inquiry would be welcome additions for a public audience.

At a time when both public services (such as those represented in the FSA-OWI data set) and funding bodies like the NEH are at risk, Photogrammar is a valuable tool to have at hand in order to make historical arguments. Highlighting the liminal spaces between research, data curation, librarianship, art history, history, and public service, Photogrammar underscores the mutable boundaries between art history–related initiatives and their impact on various fields of study. This project also makes clear the crucial role of public funding for the arts, not only for the production of art projects, but for the discoverability and usability of artifacts and records that promote the production of public and scholarly knowledge.

A. L. McMichael
Assistant Director, Lab for the Education and Advancement in Digital Research, Departments of History and Anthropology, Michigan State University