> but how odd that these should have been confined to variants of this one song, or at most two, rather than spreading further..

Like "A-wop-bop-a-loo-bop...," not odd at all. Remember that the original phrase, "dance juba," was used primarily by plantation slaves and minstrel showmen. To judge from the evidence of the databases, "jump juba" was even less common, and "jump juba ju" possible imaginary.

"Juberju" remained confined to one song and those that borrowed from it because it sounded like nonsense. Of the people who heard it, and remembered it, hardly any would have had the opportunity to use it in print, and of those, far fewer would have had the desire.

Moreover, Mick Pearce's suggestion that "jujube" is the origin is appealing. Far more music-hall audiences people must have been familiar with the jujube than with dancing juba.

The problem is that "jujube" would have to be pronounced as "joo-joo-bee" (rather than "joo-joob," the only pronunciation in the OED). That's possible - but why alter that to "joo-buh-joo"?

Well, for the rhyme, obviously. (Though rhymes for "-ee" are easy enough to find.) But if "jujube" must be mangled and given a new "meaning" unknown to the OED besides, the probability begins to shrink and shrink. Unless, of course, the author had "jujube" in mind and thought it sounded so funny he would play with it further. But there's absolutely no way to determine that.

It's fair to say that if we had only British texts, "jujube" would be a possible origin. If there were only minstrel texts, "juba" might be a better guess. As things stand, it is hard to say which is more probable - or less improbable. And whatever the answer, "juberju" itself, then and now, remains nonsensical. As a result, anyone can project any meaning they want onto it, but that's another story.

I now think any resemblance to "jubilo" is coincidental.