Rhi tested it in the staff room and it didn't work 
I cheated the autocorrect parser, it's very simple as it looks and behaves as if it only does a single pass through the word(s) matching what it finds.
If you quote my post you'll see that I inserted a couple of control characters within the word, the parser sees them as part of the word on its single pass and obviously it doesn't match to what's in the autocorrect list so it gets through. It's only after that stage that the rendering of the text is done.
So typing 's will be displayed as just 's whereas J
urgen's gets through the autocorrect and is then rendered as if correctly.
Really, to close such loophole, the autocorrect check should be performed after the rendering pass but that would likely be at a considerable cost to performance and responsiveness.
