Hi Richard
His method was more or less what most of us use when analysing ballads. However even with all of the extant versions it's almost impossible to draw cast-iron conclusions because we don't know how many versions were lost between the first and the last. We are often reduced to looking at possibles and probables. Generally the older the ballad the more it has been reworked. At least with the earliest we have, it has a lot in common with more recent Scandinavian versions.

Useful conclusions can be drawn re the evolution of a ballad if in looking at several versions over several centuries it can be shown that the rhyming steadily deteriorated. There are English ballads where this can be demonstrated.