Stardate 20010712.2006 (On Screen): NIST has developed a new form of atomic clock which is accurate to one second in fifteen billion years. This is a preposterous degree of accuracy, and in fact they've reached the point where they can measure the relativistic time compression effects which would occur simply by moving this monster across the lab and back. (Assuming they had two of them and the other sat still.) For that matter, if there are multiple instances of these beasts, they'll measure time differently as a function of the latitude and altitude that they're at, due to relativistic effects of the rotation of the Earth. Sheesh.
In fact, this clock is not as accurate as they say. What's happening here is a confusion of repeatability and accuracy. If a measurement device is very "repeatable" then if it is used to measure the same phenomenon multiple times it will provide you with a very consistent result. If it is very "accurate" then multiple instances of the device measuring the same phenomenon will give you very consistent results. Usually a measurement device will have a better level of repeatability than accuracy. Accuracy is a hell of a lot harder, and I don't think it's possible to measure time accurately to the level of error described here. In fact, I know it can't be done because it would require an absolute frame of reference, and Special Relativity says there is no such thing. (discuss)
