TESTIMONY AND AUTHORITY. 69 There are, nevertheless, some few examples of rules that might be looked upon as having been devised to attain re- liability. Thus there are rules as to the necessity in a few isolated cases for corroboration as in promise of marriage, charges made by accomplices, or in allegations of treason. There are the rules as to the ' competence ' of interested wit- nesses, or those labouring under any infirmity. There is also the rule as to ' directness ' of evidence excluding hearsay ex- cept in certain very special circumstances. Lastly, there are the rules as to the sanction under which a witness shall speak. These as will hereafter be seen are far from constituting an adequate list of safeguards. But it will be found that all except the last which has reference to penalties and, there- fore, does not here concern us readily fall into line with and find a place among the criteria of trustworthiness indicated in the paragraphs that follow. But if the Legal treatment of Testimony is inadequate, the Mathematical treatment seems positively useless. Prob- lems relating to the conflict and concurrence of testimony have commonly been regarded as belonging to the mathe- matical theory of Probability. The probability of the truth or falsity of the assertions of the several witnesses being supposed known, that of any matter which some of them affirm and others deny is by its aid deducible. The mathe- matical solution of such a problem is doubtless logically based upon the data but only when coupled with numerous restrictions expressed or implied. But these are such as to make the theoretical witness so highly abstract a per- sonage as to find no counterpart in nature, unless it be as Dr. Venn humorously puts it a bag containing black and white balls. To apply the method to real witnesses, resuming Dr. Venn's argument in the Logic of Chance, 1 would require statistics of mendacity based upon a full classification of witnesses ; and some means for identifying the class to which each witness should be assigned. It would also be necessary to assume ignoring fact that each person has a definite degree of reliability independently of the subject of testimony ; or else to base the statistics upon a classifica- tion of matter as well as of witnesses. There would still remain even then the difficulty of determining on each occasion whether the witnesses were or were not independ- ent or ever could be so absolutely ; and lastly, the surpass- ing difficulty of deciding in how many ways each witness might go wrong. 1 Chap. xvi.
