In a survey using a stratified design, let $N_h$ be the stratum size, $S_h$ the standard deviation, and $c_h$ the cost per unit. Under Neyman allocation, the sample size $n_h$ is proportional to $N_h S_h$. If the budget is constrained by $C = c_0 + \sum_{h=1}^H c_h n_h$, what is the optimal allocation $n_h$ that minimizes the variance of the population mean?