After hearing decades of arguments on the subject, I've concluded that geneology is a poor tool for studying folksongs. It's not that they aren't derived from some earlier version; it's just that the cross-breeding makes a fishnet a better model than a tree.