Georeferencing an old map in QGIS means giving a scanned historical map real-world coordinates, so it can be overlaid on modern data and compared with today's landscape. Old maps are valuable sources for studying urban growth, river change, place names, and land use, but a scanned image alone has no location. Georeferencing bridges that gap by linking points on the old map to known coordinates.

Start with a good scan: high resolution, flat, and without folds or shadows that obscure detail. Load it into QGIS's Georeferencer tool, which opens in a separate window. Your task is to find control points: features that appear on both the old map and a modern basemap, such as road intersections, church locations, river confluences, or coastline points that have not moved much over time.

Choose points spread across the whole map, not clustered in one corner, and aim for at least six to ten well-distributed points for a first attempt. More points generally improve accuracy, but bad points hurt more than extra good points help, so prioritise points you are confident about. Avoid features that may have moved, such as river banks in a shifting channel or temporary buildings.

Once you have placed control points, choose a transformation type. For a first attempt, a polynomial or thin plate spline transformation is common, depending on how distorted the old map is. Run the transformation and load the result into QGIS, then compare it with a modern basemap: do roads align, do coastlines match, do place names sit in the right places? Check the residual errors for each point, which show how far each control point moved during transformation.

If one point has a much larger error than the others, it may be misidentified or moved since the old map was made. Remove or correct it and rerun the transformation. Repeat until errors are acceptably low and the visual overlay looks plausible across the whole map. Perfect alignment is rarely possible with old maps, because surveying and printing were less precise, so aim for a useful, honest fit rather than a forced one.

Save the georeferenced map as a GeoTIFF, which embeds its coordinates, and keep a note of the control points, transformation type, and source of the original scan. That documentation lets others judge the map's reliability and reuse it with confidence. You can then overlay it on modern layers, trace old features as new vector layers, or create a swipe comparison that shows change over time.

Georeferencing old maps is slow, careful work, but it unlocks history in a form modern GIS can use. A single well-georeferenced map can reveal a lost river course, a vanished village, or a century of urban growth, all from a document that once sat flat and unlocated in an archive.

Old maps are often distorted by paper shrinkage, folding, or inaccurate surveying, so a single transformation rarely fits perfectly everywhere. Thin plate spline transformations can warp the map locally to fit control points, which helps with distorted sheets, but they can also introduce unrealistic stretching in areas with few points. Use them cautiously, and always check the result visually across the whole map, not just at control points.

For very distorted maps, consider georeferencing in sections or accepting a lower-accuracy fit for the whole sheet and noting its limits. An honest, documented approximation is more useful than a forced, precise-looking fit that misleads readers about where features actually were.

That careful, documented approach is what makes a georeferenced old map trustworthy. It may take time, but the result is a historical source that modern maps, researchers, and the public can actually use.