I was describing situations where song definitely does provide a norm. Obviously it doesn't always - but where it does, chasing after variant instrumental forms of tunes is missing the point; those different forms all radiate from a common original, maybe one too well known to be written down. This makes obsessing over oral transmission of instrumental tunes beside the point - what is being transmitted is not the tune itself, or an instrumental tradition of playing it, but rather a vocal original and a tradition of ways of rendering songs on instruments. (I'd guess that if you asked any competent Scottish danceband musician to use "Coulter's Candy" for a strathspey followed by a reel, they could do it on the fly, even though there is no tradition of it being used for either).

In a lot of traditions, the words of dance tunes survive as vocalizations done by the caller or the dancers. These may not be the original words (particularly in American old-time) but they do continue a tradition of having words for the tune.

You get a similar situation with bawdy songs. You can often detect the presence of a widely known but unprintable version, where different features of it emerge in different bowdlerizations over a couple of centuries. The bowdlerizers were all working from more or less the same original, though their results looked very different.

The bottom line is that what looks like evolution on the surface may not be when you dig a bit deeper into what's going on.