Since the number of observations in an interval is a random variable which is approximately
independent of the number in adjacent intervals, the variance of the difference between the observed
and expected number of observations is approximately the sum of the variances of the components
of the difference.