The Map Of The Thirteen Colonies

Working With Historical Colonial Maps Is A Lot More Tedious Than People Think

I spend a lot of time digitizing and georeferencing old colonial-era maps. The Map Of The Thirteen Colonies comes up constantly because people want to overlay historical boundaries onto modern GIS projects, and most of the attempts I see are sloppy. Here is how you actually do it without wasting three days on something that should take half a day. The Library of Congress has a solid set of scanned colonial maps in their geography and map division. The David Rumsey Map Collection is another reliable source, though you will want to be selective about which versions you use. Some of the popular "thirteen colonies" maps online are modern recreations drawn from secondary sources, not primary documents. That matters because they inherit whoever made them got wrong. I learned that the hard way when I tried to use a widely circulated PNG from a history blog and the Connecticut boundary was shifted roughly fourteen miles west of where it actually belonged. When I need accuracy, I go to either the LoC or Rumsey and look for maps dated between 1750 and 1780. Early maps tend to be more accurate for colonial boundaries simply because they were made closer to the period and often from actual survey data. Later maps sometimes smooth things out or combine competing claims into a single simplified version, which is useless if you need precision.

The Georeferencing Process, Actually

Most people try to georeference these maps in QGIS or ArcGIS and throw twelve control points at it using a polynomial order 2 transformation. That approach works fine for maps that are mostly intact and printed on stable paper. Colonial maps are rarely intact and rarely printed on stable paper. They warp. They have marginalia that overlaps territory. Fold lines create discontinuities. A second-order polynomial will visibly misalign coastlines and river courses even when your root mean square error looks acceptable. Here is what I do instead. I start by cleaning the scan. I remove the title cartouches, decorative borders, and any legends that sit on top of the actual map area. Those elements distort the spatial relationship and if you use a point inside a cartouche as a control, you are introducing error by definition. I also scan at 600 DPI minimum. Anything less and the coastline resolution drops enough that picking accurate control points becomes a guessing game. For the transformation itself, I use a projective transformation rather than a polynomial. Projective handles the kind of perspective distortion you get from old maps that were photographed or scanned at an angle. It also tends to be more forgiving with the warping from age. I aim for six to eight control points distributed across the map area, not clustered in one region. A common mistake is piling points along the coast because the coastline is visually distinct. That leaves the inland areas under-constrained and you end up with coastal accuracy but interior distortion that can be off by several kilometers.

My typical RMS error target is under 0.5 pixels for the control points themselves, but I also visually check the result against a modern basemap. Numbers lie sometimes. The visual check catches what the error metric misses.

A Specific Problem I Ran Into And How I Fixed It

Last year I was working on a project that required the exact colonial boundary between Virginia and North Carolina along the southern borders of the northern colonies. The map I was using showed the border as a simple horizontal line, but the actual surveyed border had meanders and adjustments that the map simplified away. When I overlaid it on modern satellite imagery, the boundary didn't match the current state line at all. The discrepancy was about two miles in places. The workaround was to stop treating the colonial map as the authoritative source for the boundary itself and use it only for general spatial context. For the actual border geometry, I pulled the surveyed boundary data from the USGS and cross-referenced it with the historical commission reports from the 1728 line survey. The colonial map told me roughly where things were. The survey records told me exactly where they were. Using both together gave me something usable. Relying on the map alone would have produced results that looked reasonable to a casual observer and were wrong to anyone who checked.

Common Pitfalls That Beginners Miss

First, people forget that colonial boundaries were often based on water features described in charters. "From the Atlantic to the Pacific" type language was common and entirely meaningless for mapping purposes. The actual boundaries were frequently disputed and negotiated decades after the original charters. Massachusetts, for instance, had a maritime boundary issue with New Hampshire that was not fully resolved until the 1820s. A map from 1760 showing the border one way and a map from 1775 showing it another way are both "correct" for their time. Picking the wrong one for your period of interest will introduce error. Second, coordinate reference systems. A lot of old maps were drawn using local surveyor frameworks, not a unified grid. Some use latitude and longitude but with inaccurate meridian measurements. Others use arbitrary grid systems specific to a single surveyor's notes. If you try to georeference using WGS84 directly without understanding what datum the original map is using, you will get shifts. I usually start with a rough NAD27 assumption for pre-1920 American maps and then refine based on the control point distribution. If my points consistently shift in one direction, that is a datum mismatch and I adjust accordingly.

Practical Output Formats

Once your georeferencing is done, export as GeoTIFF if you need to keep it as a raster layer in a GIS project. If you need vector boundaries, use the map as a tracing reference and digitize the colonial borders manually. Automated tracing on these maps produces garbage because the line weight varies, the ink bleeds, and the paper texture creates noise that vectorization tools interpret as features. Manual digitizing takes longer but produces clean lines you can actually use in analysis. I usually save the georeferencing transform parameters separately so I can reprocess the map if I find better control points later. I have lost count of how many times I came back to an old project and realized I had missed a good control point because I was rushing. Saving the intermediate work means you are not starting over when that happens.

When This Approach Fails Completely

Sometimes the source map is too distorted or too incomplete to georeference reliably. I encountered a map from 1776 that showed the colonies but had the interior regions completely blank, filled with decorative illustrations instead of geographic data. No amount of control points can fix that. In cases like this, the honest answer is to use a different source or accept that you are working with approximate boundaries at best. There is no workaround for a map that simply does not contain the information you need. If you are doing this for academic work, document your source maps and your transformation parameters. Future researchers, including your future self, will thank you when they need to verify or correct your work. I cannot count the number of times I have had to revisit a project from two years ago and reconstruct my methodology because I did not write it down properly. It is a boring habit but it prevents genuine headaches.