An upper bound on aliasing error

The aliasing error in the reconstructed frequency domain function, referred to as the foldover error, is well known. Estimates of the corresponding error in the reconstructed time domain function are apparently rarer. An upper bound on this error is derived; it depends only on that portion of the spectrum for which the frequency is larger than half the sampling frequency. Comparison with a recent bound presented by Papoulis shows the present bound to be more useful for some cases.