If a sample of $n$ observations has a sum of $S_1 = \sum x_i$ and a sum of squares $S_2 = \sum x_i^2$, derive the unbiased sample variance $s^2$ in terms of $n, S_1, \text{ and } S_2$.