Example output from Monte Carlo simulation of a random process following eq. (1) with no true orbital period derivative. The parameters used are those derived from the actual data (﹩\sigma _{\epsilon }=0.105﹩ s and ﹩\sigma _{e}=1.62﹩ s). The top panel shows the simulated mideclipse times for every eclipse, while the bottom panel shows the same data but sampled using the sampling function of the actual observational data. Note how simple fitting of short spans of data can easily detect spurious positive or negative period derivatives, even though none actually exists in this case.