Which standard of proof is best for a particular type of case? This deceptively simple question has been much discussed, but the current state of understanding is unsatisfactory. Statisticians posed a general answer; philosophers and others launched an assault on that answer; practically oriented scholars draw on both strains unsystematically; and courts generally offer little or no reasoning for their decisions. The goal of this article is to outline a systematic and complete justification for selecting one probabilistic standard of proof over another. By training a microscope on one small corner of the law - incapacity will contests - this article demonstrates the relevance of old factors, identifies several new factors, and integrates the factors into an approach that will hopefully guide future inquiry. One important implication is that the choice of proof standard will almost necessarily be tentative: too much is unknown or unknowable.