The Shape of History presents an alternative genealogy for the emergence of modern data visualization. By recovering the forgotten visualization work of the author and educator Elizabeth Palmer Peabody (1804–1894), and by recontextualizing her abstract and colorful charts in terms of contemporary ideas about visualization, we seek to challenge present assumptions about what visualization should do: “reveal” the significance of underlying data, “enhance” the clarity and legibility of data, “amplify” existing cognitive processes, or otherwise facilitate “pathways to insight.”1 Contrary to these views, Peabody intended her charts, which depict significant events in US history, to “leave scope for a little narration.”2 Her hope was that, by appealing to abstraction and affect, and by insisting that her students interpret each chart for themselves, she would encourage her students to conjure their own narratives of history, and therefore produce historical knowledge for themselves.
Peabody envisioned the process of knowledge production as interactive, although she did not employ that particular term. In addition to the charts that she included in her history textbook, which our project reimagines in digital form, she sold workbooks with sheets of blank charts. In the ideal interaction, students would read the historical narrative provided in the textbook, scan the list of events that followed, and then create their own versions of the “painted centuries,” as Peabody termed them, to serve as visual mnemonics to recall the contours of the past.3
By locating the source of knowledge in the interplay among viewer, image, and text, by insisting on multiple possible interpretations of the charts, and by elevating the sensory and affective dimensions of knowledge production, Peabody strongly anticipated several key tenets of feminist theory. Current claims about the fundamentally “situated” nature of knowledge production, the importance of embracing horizontal rather than hierarchical modes of knowledge transmission, and the irreducibility of individual acts of interpretation all find expression in Peabody's charts.4
By enhancing Peabody's charts through digital means, we seek to amplify the aspects (and their associated interactions) that align with feminist theory. In the “Explore” mode, for instance, we employ synchronized highlighting to enhance the cross-referencing between image and text that Peabody hoped to facilitate.5 In the “Lesson” mode, we employ event-by-event segmentation, combined with immediate feedback, to guide viewers through the pedagogical exercise that Peabody envisioned in her textbook. The “Play” mode evokes the experimental ethos of the workbook exercise. And in the “Compare” mode, in which we simultaneously display the same historical data in four different ways, we seek to call attention to the importance of form in terms of the meaning (or lack thereof) that any particular data visualization conveys.
The Shape of History is therefore a project about data as much as it is about visualization. The four simultaneous representations of Peabody's “data,” both independently and together, show how storing history as data results in a sense of the past that is profoundly incomplete. When visualizing this data, we can choose to present a complete picture of the past, or, following Peabody, we can take steps to prompt further exploration.
A NOTE ON DATA AND DESIGN
The Shape of History employs the lists of events included in Peabody's textbook as the source of four of the six data sets that are visualized on the site. For the sixteenth century, for example, the book contains eighteen events, listed in chronological order. Each event includes a year, a short description of the event, and a short explanation of where, and with what color, the event is represented on the chart. To make this data machine-readable, we manually created a set of CSV (comma-separated value) files, each with a row for each event. (One particularly complex event, which grouped five sub-events together, was split into five independent events.) Within each CSV file, each event was mapped to the following attributes: year, color, actor, event type, and description. The events for the sixteenth century, the project's default data set, contain twenty-three unique data objects. For the two new data sets, “100 Years of Women's History” and “100 Days Before the 2016 Election,” we created custom actor and event type values, with data points that derive from original research.
When designing the user interface, we recognized that we needed to make a choice between an interface (and related textual apparatus) that would speak to scholars and one that would speak to a more general audience. Ultimately we decided to design for the latter, as we believed this would lead to a larger audience for the project. This decision resulted in a reduction in the amount of historical and theoretical information, as well as a more tightly controlled interactions. We are considering developing a more scholarly version of the site that will allow for deeper exploration and exposition.