True when count failures in trials stay within 2.5 standard deviations of the declared permille rate (4 d^2 <= 25 n a (1000 - a), exact in integers).
count
trials
rate