Oops, on both counts. I meant split, not shift.

As for the different returns, in one case I verified that it was a unique string and so return the original, in the other I verified it was not and returned the substrings in it. However I don't keep context about locations in the original string, so I cannot tell whether a duplicate is because I found the same string through two paths. In your example through cutting on 'h' then 'e' I get to ' quick brown fox jumped ov' no matter which way I split on 'h' first.

That is fixable, but not easily.