Consider the Benjamini-Hochberg (BH) procedure for controlling the False Discovery Rate (FDR). If we have $m=20$ tests and we wish to control the FDR at level $\alpha = 0.10$, given ordered p-values $p_{(1)} \le p_{(2)} \le \dots \le p_{(20)}$, what is the condition to reject the hypothesis $H_{(j)}$?