I haven't actually seen any mentions of stylometric techniques or analyses which look into that question, although I'm sure it's been done. (The only example I can think of, the Federalist papers, was known to have been done by two people with large pre-existing corpuses to go on, and so the question was simply assigning responsibility.)
I actually have read a paper once on stylometric authorship attribution, part of which was oriented at detecting whether it was a single or multiple authors. I think "authorship attribution" were keywords in the title, but that's all I remember. So I don't have anything more to go on :)
