Would I be right in suggesting that it's not encrypted data unless someone can decrypt it to prove so? In other words, we're free to send garbled nonsense to each other.
England will imprison people who forget passwords. Here are two examples:
Also, keys which are only used for signing are specifically exempted from disclosure in the primary legislation - so don't hand over private TLS signing keys (if, and only if, you always use ephemeral keys, not static RSA) or any GnuPG/PGP signing keys (encryption subkeys on the other hand? However, GnuPG has a way to output the symmetric keys for individual messages, allowing you to limit the damage)!
Of course, if ever presented with any such order, you absolutely need very good legal advice rather than taking it from a lay random on Hacker News. I would definitely encourage you to fight, if you can.
http://en.wikipedia.org/wiki/Key_disclosure_law#United_Kingd...
.. and if there is no key then you may be in a very difficult position.
Indeed, for some schemes, even if someone cooperates fully, they will be unable to prove that they have, which could leave them in a very dangerous situation. It will also be difficult to prove that they haven't cooperated fully, but whether that is relevant depends on the type of duress they face - you may have varying degrees of success or failure facing thresholds of 'beyond reasonable doubt', 'preponderance of the evidence', or 'hammer to the kneecaps'.
A vitally important thing to know, if you're a keyholder of such a system. Given such a disadvantage, they are not very commonly used. The vast majority of all those who use (and have used) TrueCrypt don't use hidden volumes.
It's also worth pointing out that any disk usage metadata - as, for example, is kept by any and every SSD - tends to catastrophically break deniability. I don't know of anything that can do deniability with a flash device.