When computing a standard deviation of a large number of repeating values, there is a precision loss. The result is not  as it is mathematically correct, but a small number (e.g. order of 10^-10).