On Exactitude in Science