Ethan Allen2 said:I roll up my sleeves and draw. There's no other way. 😁
I looked into whether I could automate it, but the time spent trying to build and set up a program would be just as much as the time spent drawing manually.
The biggest map I ever finished was the 1895 Canada religious map.
http://www.4shared.com/photo/I7k3Jpfe/1885.html
We're talking about maybe 5,000 or 6,000 settlements there. I can't remember exactly how long it took, but it was days of work. To do that automatically, it’d still be a massive undertaking. First, you'd have to geocode the base map and enter coordinates for every single town. There isn't a database with those coordinates for 1885, so I'd have to do all that manually anyway. Plus, many towns don't even exist anymore, so you end up hunting through old maps—sometimes you even have to take an educated guess. Finding those places is a huge job in itself.
Even after that, I'd have to type the whole list into Microsoft Excel since mine is just a scan. It's an old book, so OCR wouldn't work well; that part would be manual too. All in all, just more endless hours of work.
And even if I did finish, it wouldn't really make future maps any easier. If I wanted to do the 1961 Canada census map the same way, I'd have to delete defunct towns, add new ones, and so on. And then re-type the whole list again.
Doing it by hand isn't actually that hard. Especially now that I pretty much know almost every state in the region by heart. In an hour, I can knock out 100 settlements without a problem.
Since I've done some work converting PDF to Excel files using small scripts, I can tell you it isn't a huge ordeal. However, dealing with old books—scanned images—is likely a different story. Since you have to type the names of the settlements onto the map anyway, I think it makes sense to type them directly into Excel at the start.
The biggest hurdle is adding geocoordinates for every little town. The good news is that once you determine the coordinates for a location, you can reuse them later.
The easiest part of automation is creating a function that draws borders or fills in areas. Once that's set up, you can apply it to all your maps and datasets.
The harder part is the actual creation of the map. If you can't "borrow" a base map, you have to do most of it manually. On the bright side, once a map is drawn, it remains useful for future projects. The end product is an image, ideally black and white. Georeferencing an image isn't particularly difficult.
A program designed to draw maps with borders would simply take an XML file containing the settlement data and an empty georeferenced map as inputs. The result? A finished image in just a few seconds!
The greatest advantage of automation is that it is infinitely repeatable, and those tedious, repetitive tasks are performed incredibly fast and without error.