If I must say one thing, it’s that historical mapping looked easy until I tried it. As you should be able to see, in order to make my map look even remotely accurate, I had to warp it significantly, and around 30 points of reference make it fit the modern map. I can’t imagine the number and accuracy of locations used for this kind of mapping in a serious project. The final map can be accessed via ArcGIS, QGIS, WMTS, TileJSON, and XYZ. If this historical map is accurate, it could be used to record the way land masses have changed over time, that is coastlines and other specific shape. It could also be mapped with other historical data.
As for problems with georeferencing, I was only able to identify some identical landmasses, and otherwise I had to use cities, whose centers may have varied in location over time. The accuracy of this map is not going to perfect; I don’t think that even if I added more points, it could be. Specific to this site, you can see from the link to my map that for some reason it never updated, in spite of my saving the map (Iceland especially seemed far off before I fixed it). But even with these issues, certain map formats and maps that are not historical would not fit well on this structure. We discussed, for example, mapping Narnia, but this would additionally be problematic for something like Pangea Ultima, I think, unless it was on the exact same mapped form.
As for research questions, I think that questions about historical content, like the Gettysburg map, wouldn’t necessarily be aided by mapping them on a traditional world map; instead, the data there relied more on having historical data about the land masses and troops of the time, thus comparing multiple historical maps rather than a historical and modern one.
Thanks for reading! -Kelly_H