Historical GIS
- Introduced by
- Ian Gregory and Paul Ell
- Year
- 2007
- Type
- spatial-analysis-pipeline
- Family
- spatial-historical-analysis
What does it do?
Historical GIS brings the spatial-analytic power of geographic information systems to the study of the past, building databases in which historical places, boundaries, and phenomena are tied to coordinates and to the dates at which they held.
Systematized in Ian Gregory and Paul Ell's foundational treatment, the approach addresses a problem ordinary GIS ignores: the geography of the past was not fixed. Administrative units split and merged, borders shifted, towns rose and vanished, so a historical GIS must represent geometry that changes through time. Researchers georeference old maps, digitize past boundaries, encode places in gazetteers, and link tabular historical data, censuses, tax rolls, trade figures, to these time-varying geographies. The result supports genuinely spatial questions: how phenomena were distributed, how patterns clustered or diffused, how distance and terrain shaped historical life. It operationalizes the Annales attention to geography as a force in history, letting scholars map and measure the spatial structures within which past societies acted.
The intuition behind the method
A modern map pins every place to a stable grid, but the past will not hold still on that grid. A parish absorbed by its neighbor, a county redrawn, a border moved by treaty, all mean that the same patch of ground belonged to different units at different times, and that a name in an old census may point to a territory that no longer exists. Historical GIS is the craft of making maps that remember when. It treats each boundary and place as valid only for a span of years, so that asking the system to show a region in 1801 returns a different picture than asking for 1901. Once geography carries time as a dimension, the historian can lay tabular records onto the correct historical map, then see and measure spatial patterns: where wealth concentrated, how disease spread along roads, how a frontier advanced. The intuition is simply that space mattered to people in the past as it does now, and that to study it honestly we must reconstruct space as it actually was, decade by decade.
History
Historical GIS emerged in the 1990s as historians and geographers began applying geographic information systems, originally built for contemporary planning and environmental work, to the past. Large infrastructure projects such as national historical GIS databases of changing administrative boundaries, including Britain's, demonstrated the approach's value and posed the central challenge of representing change over time. Ian Gregory and Paul Ell's 2007 Historical GIS provided the field's foundational methodological synthesis, codifying techniques for georeferencing, temporal data modeling, and spatial analysis of historical sources. The approach resonated with the Annales tradition's emphasis on geography in history. It has since grown into a broad spatial-history movement, expanding from administrative mapping toward deep maps, linked open geodata, and integration with text mining and named-entity recognition for extracting places from historical documents.
When to use it
Turn to historical GIS when your question is fundamentally spatial and your sources can be tied to places and dates: how something was distributed across territory, how a pattern clustered or diffused, or how geography conditioned historical processes. It is well suited to working with censuses, boundary records, and old maps, and indispensable when administrative units changed over the period studied, since it can reconstruct the correct geography for each date. The approach requires georeferenceable sources and consistent place identification, so it demands data with locational content and reasonable cartographic coverage. It complements longue duree analysis by giving quantitative spatial form to the geographic structures Braudel emphasized. It is unnecessary for questions with no meaningful spatial dimension or where locations cannot be reliably recovered.
Limitations and considerations
- Building time-varying spatial databases is labor-intensive and technically demanding.
- Georeferencing old maps introduces locational error that propagates into analysis.
- Changing and ambiguous historical place-names make linkage difficult and error-prone.
- Imposing precise modern geometry can give a false impression of past spatial exactness.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Reconstructs past geographies accurately by representing boundaries that change over time.
- Reveals spatial patterns and relationships invisible in purely tabular historical data.
- Integrates heterogeneous sources, maps, censuses, gazetteers, into one analyzable database.
- Gives quantitative, testable form to the geographic dimension of historical explanation.
Applications
- National historical GIS systems reconstructing changing census and administrative boundaries.
- Mapping the spatial spread of epidemics through past populations and transport networks.
- Analyzing the geography of industrialization, urban growth, and migration over time.
- Reconstructing historical trade routes and the spatial reach of markets and empires.
- Spatial studies of land use, environmental change, and settlement over the long run.
How it works
How it works
The pipeline starts by georeferencing historical cartographic sources: scanned old maps, plans, and surveys are aligned to a modern coordinate system by matching control points whose locations are known. This transformation warps the historical image so its features sit at their true geographic coordinates, allowing later digitizing and overlay. Careful georeferencing, and honest assessment of its error, is foundational, since every spatial measurement downstream inherits the accuracy of this initial alignment.
Formula
(x', y') = T(x, y), \quad T : \text{map} \to \text{coordinates}Next, features are digitized into a spatial database: boundaries become polygons, settlements become points, routes become lines, each with attributes drawn from the sources and, crucially, the date range over which it was valid. The database thus stores not just geometry and attributes but temporal extent, recording when each feature existed. This explicit encoding of validity periods is what lets the system reconstruct geography for any chosen moment.
Formula
G = \{(g_i, a_i, [t_0, t_1]_i)\}Because the past geography changed, the system represents geometry as a function of time. Boundaries are modeled so that querying a date returns the configuration in force then, handling splits, mergers, and shifts of administrative or political units. This time-varying representation distinguishes historical GIS from ordinary GIS and is essential for accuracy: comparing data across periods requires knowing that the spatial units themselves were not constant.
Formula
\text{Geometry}(t) : \text{boundaries vary with } tTabular historical data, censuses, tax assessments, vital records, trade statistics, are then linked to the spatial features by matching on place identifiers and dates. This join attaches each record to the correct historical unit for its period, reconciling the messy, changing place-names of the sources with the database's geometries. Robust linkage, often requiring careful name disambiguation, turns isolated tables and maps into an integrated spatio-temporal dataset ready for analysis.
Formula
D_i \bowtie G_j \text{ on place-key and date}Finally the integrated data are analyzed with spatial methods: mapping distributions, measuring clustering through spatial autocorrelation, modeling diffusion, and assessing how distance, terrain, and connectivity shaped outcomes. Visualization reveals patterns invisible in tables, while spatial statistics test whether observed concentrations are meaningful. This stage realizes the geographic insight at the heart of the method, treating space, as the Annales tradition urged, as an active dimension of historical explanation rather than a passive backdrop.
Formula
\hat{\rho} = f(\text{spatial autocorrelation}, \text{distance})Common pitfalls
- Ignoring boundary changes and comparing data across incompatible spatial units.
- Treating georeferenced coordinates as exact while neglecting their uncertainty.
- Mismatching records to places through unresolved name variants or duplicates.
- Reading spatial correlation as causation without controlling for confounders.
Frequently asked questions
How does historical GIS differ from ordinary GIS?
Ordinary GIS typically assumes a stable present-day geography, but the past geography changed constantly as boundaries shifted and places appeared or vanished. Historical GIS adds time as an explicit dimension, encoding the date range over which each feature was valid so the system can reconstruct the correct configuration for any chosen moment. This time-awareness is its defining methodological contribution.
What is the hardest part of building a historical GIS?
Two challenges dominate: representing geography that changes over time, including splits and mergers of administrative units, and linking tabular records to places despite inconsistent, ambiguous historical place-names. Both demand careful data modeling and disambiguation. Georeferencing old maps accurately, and honestly tracking the resulting locational error, is a further substantial and labor-intensive task underpinning everything else.
How does historical GIS connect to other digital-history methods?
It increasingly integrates with text mining and named-entity recognition, which extract place-names and events from large historical text corpora that can then be mapped. Gazetteers and linked open geodata connect these places to coordinates and time spans. Together these methods let scholars move from unstructured documents to spatially and temporally structured databases, enabling large-scale spatial analysis of textual historical sources.
Related methods
Sources
Gregory, I. N., & Ell, P. S. (2007). Historical GIS: Technologies, Methodologies, and Scholarship. Cambridge University Press.
ISBN: 9780521855631Copied.Braudel, F. (1958). Histoire et sciences sociales: La longue duree. Annales. Economies, Societes, Civilisations, 13(4), 725-753.
Open DOI referenceCopied.