Thanks, I'll check that over. I had converted the link I found, and it was mostly working, but I hadn't gotten it 100% working the way I want.

And there are at least 50 instances that need to be converted, and it will be run as part of a larger sequence, so selection is not an option. I am guessing that my best option would be running through each character of the document for a superscripted number, and continuing until I find a character that is not a number, space or comma (or is not superscripted), then processing that string.