I was talking about to use INSTR to determine if ONE letter is part of a word. Not entire words. When you test all letters of word 1 against the letters of word 2 then you will know how many times each one is contained in word 2. as long as this is one time, it's an anagram. It would be pretty insane to create an array of all potential permutations, only to compare them.
I get you - for a one word anagram it's spot on. If you want more complex results (like a bunch of sub words that are comprised of the original word's letters) then I don't think you can get away without the kind of branching permutation tree covered in the article I linked. I'm holding off on a complete solution because it seems to me that (scary) recursive functions would be appropriate. :/