Possible errors in inferring from sampled data the spectrum of a physical quantity´s fluctuations

It is shown that sampled measurements of a physical quantity can produce spectral aliasing and partially mask the real spectrum of its fluctuations with excess noise. The latter appears as a white process in both spectral and variance analysis. It is argued that, conversely, no easy conclusion on the real statistical properties of a quantity´s fluctuations can be made based on sampled measurements.