The second assumption dictates the normative
behavior of the residual error term. By assumption, the
expected value of the error term (e)
is zero. Further, OLS methods assume
that deviations from the regression line are normally distributed with a mean
of zero (0) and constant error variance.
Stated differently, the residuals
are expected to sum to zero. A normal
probability graph should show the residuals in a linear plane (slope is not
important).