The collectors who travelled into the countryside with notebooks were searching for the original version, and they did not find it, and the reason they did not find it is that a song of this kind does not have one. What they encountered in its place was that every singer presented them with something slightly different, and that the same singer presented them with something slightly different on a second visit, and they recorded these as corruptions of a text that must once have existed and had since decayed. The word corruption is the whole assumption compressed into a single term.
A song transmitted by singing is not a text that gets reproduced. It is a set of constraints -- a tune shape, a stanza pattern, a narrative with certain fixed points. Inside those constraints a singer assembles a performance out of material he has heard, forgotten, misremembered and improved. He assembles it from a stock of formulae, and that stock is shared. That is why versions collected two hundred miles apart share lines that neither singer can have obtained from the other. Nothing about this process has a first term. Asking what version is original is like asking what performance of a dance is the real one.
The collectors' method made the assumption invisible. Presented with six versions, an editor would print whichever one he considered best and relegate the remainder to a note, or he would construct a seventh from the strongest parts of all six. That is what the great published collections predominantly contain. The result reads like a text because it was made by the methods used on texts, and generations of readers have taken the printed song for the thing the singers had, when it is a thing no singer ever sang.
None of this makes the collections worthless; they preserved a great deal that would otherwise have disappeared. But the loss is specific and worth identifying. What was thrown away was the variation, and the variation was not noise around a signal. It was the evidence of how the form functioned -- what lines were fixed and what were free, what a singer would change under what conditions, how a story travelled and what happened to it on the way. A discipline that had understood the songs as performances would have collected the differences deliberately, and the material that would have been most revealing is precisely the material the method was designed to discard. There is a contemporary parallel worth observing. Recordings made from the middle of the last century onwards capture exactly what the notebooks could not, and the first thing they demonstrate is that the variation is patterned rather than random -- that a singer alters the same places in the same directions, and that the places he leaves alone are the ones carrying the story. That is a finding about form, and it was unavailable for a century because the evidence had been tidied away.
