Point‐counter analyses include two errors, determined separately by count length and by sample area. The counting error has been studied by Chayes, and his latest results admit an estimate also of the sampling error. It is postulated that the sample size and the rock coarseness do not affect this error independently; instead, they combine to determine the effective number of grainsections sampled, and this in turn controls the sampling error. The estimate of the sampling error is compared with that for the counting error in selecting the best sample sizes and count lengths for projected work.