If we look at the anchor escapement, however, we can see that it does not fully fit the definition of an ideal escapement. In an ideal escapement (and I owe a lot of this analysis to Daniels’ Watchmaking, which for a lucid explanation of the principles of a practical watch escapement is very hard to beat), impulse would be applied instantaneously at the equilibrium point, in both directions, with equal force each time in order to ensure perfect symmetry of motion (especially important in a watch). There would also be no friction involved as this dissipates energy and affects the motion of the oscillator. The anchor escapement fails on both counts – not badly, by the way; you can get excellent performance out of it – but it is not an ideal solution. Moreover, the sliding friction at the escape wheel teeth and pallets, as the curved projections of the anchor are called, requires oil, and any oil will eventually thicken and evaporate over time. The viscosity of oils will also change with temperature, and this means that the ideal escapement would be oil-free as well. A watch escapement should be self-starting – that is, its design should be such that the watch will spontaneously begin to run once a certain amount of energy is wound into the mainspring. The escapement must have good safety – that is, it should not unlock accidentally if the watch is given a shock. And overall, of course, in giving impulse and counting oscillations, the escapement should interfere with the natural harmonic motion of the oscillator as little as possible. So we have a little checklist: