Supposedly xi'i is intended to remove ambiguity in multiple subscripts. A summary reference appears at http://groups.yahoo.com/group/lojban/message/7225:
xi'i (XIhI) terminator for xi; used in mekso to disambiguate e.g. one variable
with two subscripts and a variable with a subscript that itself has a subscript.
Has this ever confused anybody in practice?
No, but an ambiguity in an obscure corner of the language is still an ambiguity and should not exist.
How is it ambiguity? If you have to subscript that high you're surpassed any listener's ability to follow, and you need to rethink the expression. Am I missing something? What about "boi"?
Elidable terminators can't exist in subscripts for technical YACC reasons.
Anyhow, it's not necessary: the syntax for multiple subscripts is xi ... ce'o ... ce'o....
But this does vaguely jog my memory (saith John Cowan) about some problem with subscripts. I will try to investigate this.