It is proposed that multiple measurements be treated as a random sequential sample for probability estimates and a measurement procedure based on a sequential plan is constructed which will ensure fulfillment of specified limits on the probabilities of type 1 and type 2 errors with a minimum sample volume.