An error of measurement as a consequence of recording the value x+ε instead of the true value x, with ε being an observation on a random variable. The random error is often assumed to have a normal distribution with mean 0 and constant variance (though this assumption should always be verified). In the case of errors introduced by rounding to a fixed number of decimal places, a uniform distribution is appropriate. See systematic error.