Of course the principles of escapements are well established. We know what is required of an escapement, but whilst it's easy to understand what's needed, it's not very easy to discover how to make it available. And I think also there's a certain amount of inhibition because one is faced with a 500 year old problem that hasn't been solved and to set about solving it without giving it sufficient consideration wasn't going to produce the results. And I realised that and... Still, I set to work and produced different escapements. It's difficult to describe these escapements, but they included the best-required principles for escapements, with the exception of the tick-tock impulse, and that was the greatest problem to find out how to introduce an impulse to the oscillator at every vibration. And finally, I produced an escapement which I called the co-axial escapement because it had two escape wheels, not side by side as with the earlier watch, but one on top of the other, and this gave the tick-tock impulse and didn't require lubrication. So, as far as I could see, the problem was solved and it only meant putting it to test to find out how it went. And I tested it during a period of 20 years, not because it wanted all that testing, but because I couldn't find anyone in the industry to take an interest in it. But during those 20 years, it proved itself to be a superlative timekeeper and some examples have run for the best part of a year without showing any variation in timekeeping. So obviously it was suited to its task. It didn't need the lubrication so it wasn't affected by a change in temperature.