Thanks a lot BOAC for your explanations.
I still find it fascinating that a pure mechanical device would drift as little as 30 nm after an oceanic crossing of say 3000nm, so a 1% error.
Fortunately (or unfortunately?), EFIS is becoming more and more straight-forward for GA aircraft. Heck the last prop I flew (a C172) had a G1000 cockpit - compare that to a B733/B757/B767 cockpit and you'll get pretty envious if you're into EFIS