The problem of our title is not usually considered as purely scientific, but as pertaining largely to philosophy and metaphysics. This means that the involvement of the scientist is partly emotional and partly rational. The need of the investigator to control or counteract his own, often unconscious tendencies is the most outstanding feature of our subject. On recognizing this, it becomes indispenable to find a purely formal criterion to which one can refer time and again to keep the intrusion of such nonrational tendencies in check.