What I tend to do is to wake about five in the morning-this happens quite often-think about the invention, and then image it in my mind in 3D, as a kind of construct. Then I do experiments with the image...sort of rotate it, and say, 'Well what'll happen if one does this?' And by the time I get up for breakfast I can usually go to the bench and make a string and sealing wax model that works straight off, because I've done most of the experiments already.