The reasoning of this passage (in which the uniformity of the laws of nature is assumed) seems to me a good illustration of what must be considered the principle or form of an induction. The conclusion, which is its scope, is, by its own confession, not proved; but it ought to be proved, or is as good as proved, and a man would be irrational who did not take it to be virtually proved; first, because the imperfections in the proof arise out of its subject-matter and the nature of the case, so that it _is_ proved _interpretativè_; and next, because in the same degree in which these faults in the subject-matter are overcome here or there, are the involved imperfections here or there of the proof remedied; and further, because, when the conclusion is assumed as an hypothesis, it throws light upon a multitude of collateral facts, accounting for them, and uniting them together in one whole. Consistency is not always the guarantee of truth; but there may be a consistency in a theory so variously tried and exemplified as to lead to belief in it, as reasonably as a witness in a court of law may, after a severe cross-examination, satisfy and assure judge, jury, and the whole court, of his simple veracity. 2. And from the courts of law shall my second illustration be taken. A learned writer says, “In criminal prosecutions, the circumstantial evidence should be such, as to produce nearly the same degree of certainty as that which arises from direct testimony, and to exclude a rational probability of innocence.(24)” By degrees of certainty he seems to mean, together with many other writers, degrees of proof, or approximations towards proof, and not certitude, as a state of mind; and he says that no one should be pronounced guilty on evidence which is not equivalent in weight to direct testimony. So far is clear; but what is meant by the expression “_rational_ probability”? for there can be no probability but what is rational. I consider that the “exclusion of a rational probability” means “the exclusion of any argument in the man’s favour which has a rational claim to be called probable,” or rather, “the rational exclusion of any supposition that he is innocent;” and “rational” is used in contradistinction to argumentative, and means “resting on implicit reasons,” such as we feel, indeed, but which for some cause or other, because they are too subtle or too circuitous, we cannot put into words so as to satisfy logic. If this is a correct account of his meaning, he says that the evidence against a criminal, in order to be decisive of his guilt, to the satisfaction of our conscience, must bear with it, along with the palpable arguments for that guilt, such a reasonableness, or body of implicit reasons for it in addition, as may exclude any probability, really such, that he is not guilty,—that is, it must be an evidence free from anything obscure, suspicious, unnatural, or defective, such as (in the judgment of a prudent man) to hinder that summation or