> Is the Freebase data good enough for your purposes? It appears that it lists the
> most important characters first, but that may just be the order in which they
> were added. You may not be able to rely on that sequence.
> A Tale of Two Cities: http://www.freebase.com/m/09c55p
> Pride and Prejudice: http://www.freebase.com/m/060xy
Even if a single list or database is not compleat, one could cascade down to the next option: if 650a is blank in x, then gather from y, and so on down the line until *something* is in the field. Since this is for "Classics" one assumes that they are widely held, at least according to some genres and some data from da OCLC.  Indeed, if one were really trying to show off a visualisation, were I they, I'd hit those widely held lists since frequency would have those titles on most people's shelves.
The chore to me is rather finite and rather well defined. It's just a lot of yeoman's work, but the bulk of that can be done with a bot or army of them. It would be much easier if the list requested by the petitioner were well defined somehow. At that stage it would be a matter of grabbing ISBNs,cascading scripts, and deciding on outputs. It might even be a simple definition of Classics as predating a certain publication year.
The list's the thing!
http://www.oclc.org/research/wtworldcat/toplovestories.html?urlm=168868(The murky depths of my memories say that at some point, Bleak House was top dawg. I've tested on that assumption since.)