Layering in GIS as a Method of Historical Deconstruction and Source Criticism
By Julius Wilm
Quantitative approaches have fallen out of fashion in the discipline of history. If one goes back a few decades, this is a surprise. In the 1960s to 1990s, quantification was hailed as a key to the renewal of the discipline, which would finally satisfy scholarly criteria in representing historical reality on a large scale instead of relying on anecdotes. Researchers harnessed enormous economic and demographic datasets to address structural dimensions of historical processes in new ways. For a while, econometrics and social science methods seemed to complement and, in some cases, even replace historical hermeneutics.
In 2023, quantitative approaches no longer have a preeminent standing within the mainstream of historical research. In part, this can be attributed to the fact that quantification seems rather ill-suited for the questions discussed in the discipline today. The New Cultural History’s analyses of power and an increased consideration of marginalized populations have made the data positivism of quantitative history seem increasingly questionable. Instead of an unproblematic presentation of the big picture and giving insight into the mechanisms of the historical formations under study, it is argued that historical statistics primarily depict the perspective of the apparatuses that generated them. Governments and businesses created most statistics, and historical studies that base themselves too firmly on these outputs run the risk of reproducing the perspectives of these actors.
While today one finds few statistics and even fewer equations in historical studies, there is, at the same time, a modest revival of statistical methods coming from the digital humanities and digital history. Due to the increasing availability of statistically processable historical sources and data, there is a renewed experimentation with making large-scale data analysis useful for historical scholarship. For now, however, the new digital experiments have hardly been able to soften the reservations raised against their quantitative predecessors. Moreover, many digital projects have been designed to visualize sources in new and exciting ways rather than to develop new arguments in the discipline of history. Quantitative methods—digital or otherwise—are rare now in history.
This blog post discusses how georeferencing of sources in GIS can break with the data positivism of earlier quantitative approaches. The method of layering different sources allows us to represent the one-sidedness of certain groups of sources and thus develop a historical representation that transcends their inherent bias. Using open-source software such as QGIS, one can layer data to take advantage of the strength of these models—namely their scale and ability to distinguish typical from atypical developments—while also reflecting on the one-sidedness of data.
Although this text aims to show how a digital method can be fruitful for historians, it argues that one should read data from an understanding of historical source criticism, more than is usual in the data sciences. For historical source criticism, the limitations and one-sidedness of sources are a fundamental assumption—and it also knows ways of dealing with them. Strangely, historians sometimes detach themselves from these insights when dealing with quantitative data (particularly in digital models) to the advantage of a positivistic understanding.
I draw on a Postdoc project that I am working on at Leipzig University’s SFB 1199. This project explores the overlapping logics of imperial and national-democratic logics of spatialization between the US Civil War and the Great Depression. It does this through an in-depth study of how land appropriation by white settlers under the U.S. Homestead Act of 1862 contributed to the dispossession and attempted destruction of Indigenous nations between the 1860s and the 1920s. Like many other works on the history of marginalization, it deals with a very uneven source situation—almost all records and, even more so, measurable quantitative data have been produced by different U.S. government institutions. Indigenous actors, whom the project pays special attention to, have left few statistical or comparable sources. This blog post discusses how the method of layering in GIS can address this problem in a new way.
The principal sources I am layering in GIS are:
– Statistics on homestead land claims, that is, applications by settlers for free land from the U.S. government at 236 local land offices between 1863 and 1912. The statistics were compiled by the Accounting Division of the U.S. General Land Office, and they cover original applications, finished applications, and early purchased (“commuted”) land titles.
– Maps of Indigenous land cessions and Indigenous reservations. These maps were compiled in a project of the Smithsonian Institution, based on official land cession treaties by the U.S. government.
– Data on frontier clashes between Indigenous nations, U.S. Army personnel, and civilians. These are based primarily on army data.
In the following, the text presents some problems of these sources and shows why layering in GIS makes them useful for critical analysis.
How was historical data handed down?
In developing this project, the first problem was how historical settlement data was handed down. The “Historical Statistics of the United States” only offers settlement data on the national level, making any analysis of local developments impossible.  And on closer examination, the most detailed dataset available—published in a 1962 Bureau of Land Management brochure and used frequently in the literature—turned out to be highly questionable. While these figures are used widely, particularly in economic history, the data is limited, as it gives:
- Finished homestead claims, instead of the number of original filings, finished claims and early purchased parcels as recorded by the U.S. General Land Office
- Data on the state level, instead of the smaller land districts
In two ways, the data is simply incorrect, as it:
- Inflates all the data for 1906 and onwards. The post-1905 figures include early purchased (commuted) lands as finished homesteads, while for the previous years they were not counted under this category.
- Conflates fiscal years and calendar years. (Fiscal years started on July 1 of the previous calendar year.)
To access the original data collected by the U.S. General Land Office, it had to be transcribed from the original statistical volumes kept at the National Archives in Washington, DC.
The data was subsequently georeferenced with the coordinates of specific land offices. Furthermore, in cooperation with Robert K. Nelson and Justin Madron of the University of Richmond’s Digital Scholarship Lab, an atlas of land districts that allows showing the land office data in polygon maps was also digitized. An interactive map based on the data discussed here was published last year in the Lab’s digital history atlas American Panorama.
How was historical data generated?
As the land office statistics show, just visualizing the data that a government collected can be less straightforward than one might expect and can involve more labor than downloading a ready-made dataset. However, there are also problems with more accurate digitizations of government-created data. And these problems are inherent to their original function. Two examples from the project:
One important component are maps of Indigenous land cessions and reservation boundaries. These maps were created by Charles Royce for the Smithsonian Institution in the early 20th century and digitized in 2018 by the Forest Service and combined with data spreadsheets by historian Claudio Saunt.
Mapping Indigenous land cessions may seem relatively straightforward: The U.S. West was ceded piece by piece by its Native owners under forced treaties to the United States government. A specific date demarks each forced transfer. However, the historical reality was a lot more complex than this: The U.S. government frequently made treaties with just some leaders of an Indigenous group. If resistance to removal was too significant, the government would often try to get more leaders on board in additional treaties. Government negotiators were also notorious for getting groups to sign by adding later ignored stipulations or giving verbal promises that were not in the written text.
Furthermore, beginning in 1871, “land cessions” were based on unilateral declarations of the U.S. president. The map here reduces this complicated process into a binary distinction of areas that are either still Indigenous territory or have definitively been acquired by the U.S. government. What this digital map visualizes is the viewpoint of the U.S. government, which was very much interested in having definitively settled—or “quieted” in the terminology of the day—Indigenous claims. The actual process of dispossession was a lot messier and far less straightforward to map. Some areas were officially ceded three and more times by treaty, after which some Indigenous leaders still held out. However, the map only allows for one definite date, and it imposes the construct of exclusive non-overlapping claims on far more complex land use patterns by Indigenous nations.
A second component is georeferenced data on frontier clashes between the 1860s and 1890 involving Indigenous people in the public land states (excluding Texas) that I transcribed from two books by Gregory Michno.
This dataset of frontier clashes radically decontextualizes and flattens the differences between violent eruptions. Reduced to timed point data with a spreadsheet of information on Indigenous nations involved, army or U.S. civilians, and casualty figures for both sides, these massacres, raids, or battles appear to be fundamentally the same. Devoid of their context and the specificity of events, the 1864 Sand Creek Massacre in eastern Colorado and the 1876 Battle of Little Big Horn in Montana look similar. However, one was a genocidal massacre of villagers, and the other was the severe defeat of a military unit in a war of displacement. The reduction to numbers suggests a moral equivalency between the combatants and understanding the Indian wars as one big competition. The origin and the process by which these figures were curated also get lost once they are mapped in this way. We mostly get a view based on the U.S. military’s reports, which were limited in their insight and were frequently self-serving to the officers who wrote them. However, this is not apparent once battles have become positivist data visualized on a map.
Why is it still valid for a critical history to use these datasets? On the one hand, the weaknesses and biases of the data sets can undoubtedly be reflected—and one is not simply captive to them in the research design. The data requires explanation and context that can be given in texts. On the other hand, the data can also be deconstructed by layering—which gives the impressive scale of the data layer and, simultaneously, allows the historian to overcome and contextualize the data’s one-sidedness.
How does source criticism work in GIS?
How does this work? Why would combining individually defective components in a map create new insights?
In this map, one can see frontier clash data and the land cessions and reservations maps layered. The “clashes” are now embedded in the context of a struggle for control of western lands. Many took place in areas that were still the sovereign domain of Indigenous nations, even by the U.S. government’s definition, so though army reports may classify them as attacks by Indigenous people, they are part of an ongoing invasion. The layering also shows significant overlaps in claims that the land cessions maps suggested did not exist. There are numerous clashes in areas that, according to the official map of the U.S. government, had passed into its ownership. The Arapaho and Cheyenne in Kansas, Nebraska, and Colorado, the Paiute in Oregon, and the Lakota in Montana, for example, fought for their lands long after Washington decision-makers had deleted their titles on U.S. government maps.
While turning sources into data is problem-ridden, combining these data layers shows their limitations. Critical contextualization and deconstruction allow us to make the sources represented by these data layers speak in new ways. The enormous scale of these visualizations allows us to trace patterns that in-depth case studies lack.
How does data layering provide concrete insights into the historical relationship between homesteading and Indigenous dispossession? Here are some preliminary findings regarding this question.
When U.S. Congress passed the Homestead Act in May 1862, most of the area where homesteaders in the following five decades would claim land was still “Indian Country,” even according to the U.S. government’s legal understanding. The fact that these lands became accessible to settlers was a clear result of the expulsion of the previous Indigenous inhabitants. The data supports the general statement that homesteading was premised on ongoing Indigenous land dispossession.
The map also documents a significant overlap between the frontier wars and homesteading. In theory, and according to the law, the U.S. government should already have become the sovereign owner of land that homesteaders could file claims for. However, as this detailed map of Nebraska, Kansas, and Colorado shows, there was still extensive fighting when homesteaders started filing for land in these states.
Going back to the national level, there is a synchronous history of homesteading and the diminishment and abolishment of unceded Indigenous lands and reservations throughout the Great Plains and Far Western states from the 1860s to 1880s. The pattern is not equally recognizable in all states and territories, as sometimes dispossession and wars preceded the arrival of larger groups of homesteaders by many years.
While there is a general link between land dispossession and homesteading in the 1860s to 80s, the connection is not equally strong in all areas. Moreover, it is not possible to isolate homesteaders as the specific driver of dispossession. Miners, loggers, cattle drivers, land speculators, the army, and other white interests may have been implicated equally.
Starting in the 1890s, land dispossession took on a different form of so-called surplus land sales on reservations opened under the Dawes Allotment Act of 1887. Some of these so-called “openings” were later used to push for reservations’ abolishment as separate political entities. On the animated map, reservations change their color from a dark purple to a more transparent tone when that happened. These sales diminished and even hollowed out the Indigenous and tribal land base on reservations that remained in place or were later restored. Therefore, the visualization of these data layers suggests a strong continuity of Indigenous marginalization between the different programs by which western territories were turned into real estate. These reservation openings are visualized in this map.
Transregional historiography often has to confront the problem of wanting to reflect the agency of colonized or other marginalized populations, while sources are very one-sided. Most sources—and especially quantitative data—mainly reflect the views and motives of powerful institutions. Layering data in GIS can at least address this problem in a new way by juxtaposing data and creating models of contextualization.
Geospatial visualizations of layered source data are hardly a finished history. But they provide a powerful heuristic entry point to historical processes and suggest what specific context and sources could be relevant. Research in primary archival source materials is necessary to flesh out the processes that layered visualizations only suggest. First-hand testimonials add human voices and differing perspectives. These materials can then, in turn, be contextualized within the layered geospatial data.
For my project, the layering of quantitative data in GIS has been instrumental in getting beyond the limited and self-interested perspective of the data’s creators. Layering is a tool that the previous quantitative history of the 1960s to 1990s did not have—which may, in part, have underwritten this tradition’s uncritical use of data.
Even today, it may not be possible to use this method of data collation on just any historical subject, given the spotty availability of historical datasets. More geospatial historical data from finished projects is becoming available all the time. And it is worthwhile to think about and experiment with how modeling, layering, and contextualizing historical data in GIS can be developed further in tandem with historical source criticism. Both data science and history can be further refined by learning from each other.
 Michelle L. Butler and Myron P. Gutmann, “Vacant lands and disposal of public lands: 1802–2001.” Table Cf71-78 in Historical Statistics of the United States, Earliest Times to the Present: Millennial Edition, edited by Susan B. Carter, Scott Sigmund Gartner, Michael R. Haines, Alan L. Olmstead, Richard Sutch, and Gavin Wright. New York: Cambridge University Press, 2006. http://dx.doi.org/10.1017/ISBN-9780511132971.Cf1-167
 Homesteads (Washington DC: U.S. Bureau of Land Management, 1962).
 See U.S. General Land Office. Accounting Division, “Annual Statements of the Disposal of Public and Indian Lands” (1885-1925), R.G. 49, UD 790, National Archives I, Washington, DC; U.S. General Land Office. Accounting Division, “Quarterly Records of Disposal of the Public Lands” (1883-1909), R.G. 49, UD 789, National Archives I, Washington, DC; U.S. General Land Office. Accounting Division, “Records of Disposal of Public Lands under the Homestead Laws” (1863-1912), R.G. 49, UD 788, National Archives I, Washington, DC.
 Charles C. Royce, “Indian Land Cessions in the United States,” Eighteenth Annual Report of the Bureau of American Ethnology, 1896-97, 1902, 521–964; U.S. Department of Agriculture. Forest Service, “Tribal Land Cessions in the United States,” April 19, 2018, https://data.fs.usda.gov/geodata/edw/edw_resources/meta/S_USA.TRIBALCEDEDLANDS.xml; U.S. Department of Agriculture. Forest Service and Claudio Saunt, “Tribal Ceded Lands Table,” April 19, 2018, https://data.fs.usda.gov/geodata/edw/edw_resources/fc/S_USA.TribalCededLandsTable.gdb.zip; Claudio Saunt, “The Invasion of America,” July 31, 2017, http://www.ehistory.org/projects/invasion-of-america.html;
 Gregory Michno, Encyclopedia of Indian Wars: Western Battles and Skirmishes 1850-1890 (Missoula, MT: Mountain Press Publishing Company, 2003); Gregory Michno and Susan J. Michno, Forgotten Fights: Little-Known Raids and Skirmishes on the Frontier, 1823 to 1890 (Missoula, MT: Mountain Press Publishing Company, 2008).
Julius Wilm is a postdoctoral researcher at Leipzig University’s Collaborative Research Centre (SFB) 1199 “Processes of Spatialization under the Global Condition.” He obtained his PhD in Anglo-American History from the University of Cologne with a dissertation on free land colonization schemes in the antebellum United States and has taught at the universities of Copenhagen and Lucerne. In 2019–2020 he was the Gerda Henkel Postdoctoral Fellow in Digital History at the German Historical Institute Washington and George Mason University’s Roy Rosenzweig Center for History and New Media, where he began work on a digital mapping project on the Homestead Act with a particular emphasis on the law’s impact on Native nations throughout the US West between 1863 and 1912.
Further articles in the Digital Humanities Interface series on TRAFO:
Katarina Ristić and Nikola Ristić, Web Scraping and Digital Archives: A Program for the Retrieval of the Transcripts of the
International Criminal Tribunal for Former Yugoslavia, 14 September 2022
Elton Barker, Semantic Geo-Annotation for Ancient History and Beyond, 22 June 2022
Diana Roig-Sanz, Towards a Truly Global Digital Humanities, 20 April 2022
Susan Grunewald, Ruth Mostern, and Karl Grossner, The World Historical Gazetteer: A Digital Humanities Interface for Transregional Research, 21 February 2022
Thorben Pelzer, Retracing Professional Mobility: Historical Network Analysis through CERD, 11 November 2021
Ninja Steinbach-Hüther and Thomas Efer, The Digital Humanities Interface – An Introduction, 28 September 2021
Citation: Julius Wilm, Layering in GIS as a Method of Historical Deconstruction and Source Criticism, in: TRAFO – Blog for Transregional Research, 13.04.2022, https://trafo.hypotheses.org/46489
Diesen Blogbeitrag zitieren
Forum Transregionale Studien (2023, 13. April). Layering in GIS as a Method of Historical Deconstruction and Source Criticism. TRAFO – Blog for Transregional Research. Abgerufen am 25. Februar 2024, von https://doi.org/10.58079/ut6o