A two-stage design for choosing among several normal treatments in comparison with a control: The case of common, known variance