Suppose you characterized the variation in the weight of tablets from a new process and then wanted to use that work to estimate the variation in weight that could be expected in a multi-tablet package.
One way to get the multi-tablet estimate is to simulate a large number of tablets with a specified mean and sigma, using a normal distribution. Then divide the individual tablets into a large number of subgroups, sum the weights and then look at the variation in the subgroup
between and within variation.jsl