The fact that the scientific investigator works 50 percent of his time by nonrational means is, it seems, quite insufficiently recognized.