In a survey using a stratified design, the population $N=1000$ is partitioned into two strata $N_1=800$ and $N_2=200$. We want to estimate the population mean with a sample $n=100$. If the stratum variances are $S_1^2=16$ and $S_2^2=64$, what is the optimal sample size $n_1$ under Neyman allocation?