Probability and the Law

8/4/2019 Probability and the Law

1/18

California Law Review, Inc.

Relevancy, Probability and the LawAuthor(s): George F. JamesSource: California Law Review, Vol. 29, No. 6 (Sep., 1941), pp. 689-705Published by: California Law Review, Inc.Stable URL: http://www.jstor.org/stable/3477428

Accessed: 04/05/2009 10:53

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=clr.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the

scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that

promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

California Law Review, Inc. is collaborating with JSTOR to digitize, preserve and extend access to California

Law Review.

http://www.jstor.org
http://www.jstor.org/stable/3477428?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=clrhttp://www.jstor.org/action/showPublisher?publisherCode=clrhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3477428?origin=JSTOR-pdf


2/18

Relevancy,Probabilitynd the LawGeorgeF. James*

SINCEcholarsfirst attemptedto treat the commonlaw of evidenceas a rationalsystem, relevancyhas been recognizedas a basic con-cept underlyingall further discussion. Thayer gave this recognitionits classic form:

"There s a principle-not so mucha ruleof evidenceas a pre-suppositionnvolved n the very conception f a rational ystemofevidence, s contrastedwith the old formalandmechanicalystems-which forbidsreceiving nything rrelevant, ot logicallyproba-tive."1"The twoleadingprincipleshouldbe broughtnto conspicuousrelief,(1) thatnothings to be receivedwhich s not logicallyproba-tive of somematterrequiringo beproved;and (2) thateverythingwhich is thus probative houldcomein, unlessa cleargroundofpolicyor lawexcludes t."2The proposedCodeof Evidence of the AmericanLaw Institute (here-inafter referred to as the Code) substantially follows Thayer.3

Early attempts to state the requirementsof relevancywere con-fused and fragmentary,but in the suggestedrules, and even more inthe suggestiveillustrations,the germof moderndoctrine-both soundand fallacious-may be found.Thus Phillipps,writingin 1814,statedfive primary general rules governingthe testimony of witnesses, ofwhich the first was that "Evidence must be confinedto the points inissue".4The author then expandedand illustrated this general rule,saying in part

"And, irst,as the intentof evidence s to ascertainhe truthoftheseveraldisputedacts orpoints n issue,eitherononesideor theother,no evidenceoughtto be admitted o any otherpoint.Thus,*Associate Professor of Law and Assistant Dean, University of Chicago Law School;

Ph.B., J.D., University of Chicago; LL.M., Columbia University. Author of articles invarious legal periodicals.1THAYER,A PRELIMNARY REATISE N EvmENCEAT THECOMMONLAW (1898)264. 2 Ibid. at 530.a " 'Relevant evidence' means evidence having probative value upon any materialmatter and includes opinion evidence and hearsay evidence." "'Material matter' meansa matter the existence or non-existence of which is provable in the action." "... allrelevant evidence is admissable." CODEOPEvmrENCE. No. 2 (Am. L. Inst. 1941) Rules

1(10), 1(7), 9(f).4 PmHL PPS,EVIDENCE1814) 69.


3/18

CALIFORNIA LAW REVIEWforexample, n an action of debt upona bond,if the defendantpleadsnon est factum, (which puts in issue whether it be the defendant'sdeed or not), he cannot give a releasein evidence.Nor, in an actionof trespass for an assault and battery, will the defendant be allowedto prove, under the generalissue, that he was first assaulted by theplaintiff.For the same reason, it would not be allowableto shew, onthe trial of an indictment,that the prisonerhas a general dispositionto commit the same kind of offence as that charged against him.Thus, in a prosecutionfor an infamous crime, an admissionby theprisonerthat he had committedsuch an offenceat another time andwith another person, and that he had a tendency to such practices,ought not to be admitted.So, when a right is claimedby custom in aparish, proof of a similar usage in an adjoining parish, is not evi-dence of the custom. But where all the manors within a certain dis-trict are held by the same peculiar tenure, and a question arises, inany one of the manors,uponan incident to the tenure,evidencemaybe given of the usage which prevails in any of the other manorswithin the district."5Phillipps' illustrations bring out a double meaning in the require-ment of relevancy which continues in the modern statements and tendsto produce confusion in the cases. Relevancy, as the word itself indi-

cates, is not an inherent characteristic of any item of evidence butexists as a relation between an item of evidence and a propositionsought to be proved. If an item of evidence tends to prove or to dis-prove any proposition, it is relevant to that proposition. If the proposi-tion itself is one provable in the case at bar, or if it in turn forms afurther link in a chain of proof the final proposition of which is prov-able in the case at bar, then the offered item of evidence has probativevalue in the case. Whether the immediate or ultimate propositionsought to be proved is provable in the case at bar is determined bythe pleadings, by the procedural rules applicable thereto, and by thesubstantive law governing the case. Whether the offered item of evi-dence tends to prove the proposition at which it is ultimately aimeddepends upon other factors, shortly to be considered. But becauserelevancy, as used by Thayer and in the Code, means tendency toprove a proposition properly provable in the case, an offered item of

5Ibid. Peake, on whose text Phillippsgreatly relied,had said in 1801,"Anotherruleis, that the evidencemust be appliedto the particular act in dispute;.and there-fore no evidencenot relating o the issue,or in some mannerconnectedwith it, can bereceived;nor can the characterof eitherparty to a civil causebe called in question,unlessput in issueby the very proceedingtself, for every cause s to be decidedon itsown circumstances,nd not to be prejudicedby any matterforeignto it." PEAKE,ACOMPENDIUMOFTHELAWOFEVIDENCE(4th ed. 1813) 7.

[Vol.29690


4/18

RELEVANCY, PROBABILITY AND THE LAWevidencemay be excludedas "irrelevant" or either of these two quitedistinct reasons: because it is not probative of the proposition atwhichit is directed,or becausethat propositionis not provablein thecase.6

The distinction between these two senses of irrelevance seemsnot to have been observed by the early authors.Thus Phillips com-bines in a single paragrapha rule againstevidenceof a releaseundera plea of non est factum with a rule against provingthe customs ofone parishby evidenceof the customs in an adjoiningone, and evenstates that rulesagainst proof of self defense underthe generalissuein an assault and against showingthe generalcriminaldispositionofan accused exist "for the same reason".

Only lack of careful analysis-an easy satisfaction with surfacesimilarities-can explain such statements as Phillipps'. A little con-siderationshows that in the bond case, for example, no one wouldsupposeor urgethat proofof a releaseis in any sense probativeof thepropositionthat the defendant never executed the bond sued upon.This evidence must have been offeredin the erroneous belief that itconstituted an ultimate material fact available as a defense. On theotherhand, it is equallyclear that proof or disproofof the customofparishB would not be material-in the sense of involvingan ultimatematerialfact-in a suit broughtor defended on the custom of parishA. In such a case the customin B could be useful only if relevant,asgiving the basis for a permissible inference as to the custom in A.Phillipps' appreciationof the form of the argument in the secondcase, togetherwith a basic understandingof the "methodsof agree-ment and difference",7appears from his approval of a third case,ruling that usage in one manormay be proved to show the incidentsof tenure in another manorin the district providedthe two are heldby the samepeculiartenure.Nevertheless,he places these threecasestogetherwithoutnotingthat in the first we are dealingwith a problemof pleading, while genuine questions of evidence arise in the secondand third.

The same confusion can exist between questionsof evidence and6 Henceforth, for brevity, propositions of ultimate fact properly provable in a caseunder the pleadings and substantive law will be referred to as "material propositions";the characterization "relevant" will be reserved for propositions of evidence which,whether or not material themselves, tend to prove other propositions which are material.

"Irrelevant" and "immaterial" are of course the contraries of the two first terms, asdefined.7 See 1 WIGMORE,VIDENCE3d ed. 1940) ? 12.

1941] 691


5/18

CALIFORNIALAWREVIEWof substantive law, as Phillipps proceeds to demonstrateby accord-ing uniformand consolidated treatmentto cases holding that a civildefendant cannotprovehis goodcharacter,even to rebutimputationsof fraud, that in an action for criminal conversationthe defendantmay prove the light characterof the plaintiff'swife (in mitigationofdamages), and that in an action for malicious prosecutionthe de-fendant, after circumstancesof suspicion, may show the plaintiff'sgenerally bad character as evidence of reasonablesuspicion.Under moderncodes the substantive law confusion is more com-mon than the one basedon rulesof pleading,and may appearin verysubtle forms. Let us analyze a single interesting case, Union Paint& VarnishCo. v. Dean,8an action of assumpsit to recover the pur-chase price of waterproofroof paint. The defendantrelied upon theplaintiff'swarrantythat the paint would wear for ten years, breachof which he sought to show by proof that another drum of paint ofthe same brand,whichhe had purchasedsix monthsearlier,not onlyhad failed to prevent leaks but had ruined the shingles to which ithad been applied. The drumof paint in issue, purchasedjust beforeleaks developed in the first roof painted, had never been opened.Reversing the trial court, the SupremeCourt of Rhode Island heldthat the defendant's offer of proof (apparentlyalmost the only evi-dence offeredin defense) should have been received,saying:

"Ifpaintof thesamebrand, oldby the sameconcernunder hesamewarrantywithinsixmonths,hadprovedwithin hattime to benot inconformitywiththewarranty,n that t was notonlynotsuit-ableforstopping ndpreventingeaksbutwasactually njuriousoa roof,a personmightwell hesitatebeforeusingmorepaintof thesamebrandwhenhe hadno reason o expect hesecond ot to beanybetterthan the first."9Consideredas evidence of the condition of the second drum of

paint, proof of the results of use of the first is not very impressive.Waivingany doubtswhetherthe leaks in the first roof weretraceableto defects in paint in the first drum,there is no showingwhetherthedefects in the first drumof paint were due to poor ingredients,to apoor formula, or to some error in preparation.If poor ingredientshad been used, there is no showingthat use of poor ingredientswasa policy of, ratherthan an errorof, the plaintiffcompany.It is easierto believe that one lot of defective paint went out than it is to believe

8 (1927) 48 R. I. 288, 137 Atl. 469; MORGANand MAOImRE,ASESON EvmDENCE(1937reprint)289.9 Ibid. at 291, 137 Atl. at 470.

692 [Vol.29


6/18

1941] RELEVANCY, PROBABILITY AND THE LAWthat plaintiffcustomarilysold,undera ten-yearguaranty,waterproofroofpaint whichwouldrotout shinglesand cause leaks in six months.And the twodrumsof paintwereprobablynot out of onelot; certainlythere was no showing that they were. Proof of the condition of thepaint in the first drum was of negligible value in judging the prob-able characterof the paint in the second, unopeneddrum.It merelyshowed that plaintiffcompanysometimessold bad paint. If the issuewas whether the paint in the second drumwas bad, an issue on whichthe defendant had the burden, the trial judge's ruling seems sound.At worst, the issue is close enoughso that an appellatecourt shouldnot reverse. Yet there is still a ring of reason to the supremecourt'sstatement that "a personmight well hesitate before using morepaintof the same brand".He would hesitate to risk ruininga second roofeven if he only feared that the second drum of paint might be nobetter thanthe first.Andif he was reasonable n his hesitation,shouldthe plaintiffbe allowedto recovereven if it couldshow at trial that thepaint in the second drumwas perfectly good? The defendant,reason-ably hesitant to use the doubtfulpaint, by now has probablypaintedall of his roofs with some other paint and has no furtheruse for thedrumwhichhe is tenderingback to the vendor.If the customer is tobe protected,even against proof that the second drum of paint wasin fact satisfactory (as the writershouldlike to do in such a case10),a novel rule of substantive law stands revealed behind a somewhatdoubtfulrulingon evidence.But after excludingall cases which turn upon the materialityorimmaterialityunder the pleadings and substantive law of ultimatepropositionssought to be proved,there remainmany cases in whichthere is no question of the materialityof the propositionsought tobe provedand the probativevalue of the offeredevidence is the realissue. These cases, and these alone, raise the problemof relevancyasa problem n the law of evidence. How shouldthey be handled?Thayer, after stating the principlewhich forbids receivingany-thing not "logicallyprobative",excludedlegal criteria from furtheroperation,saying

"Howarewe to knowwhatthese forbiddenhingsare? Not byany ruleof law. The lawfurnishes o test of relevancy.Forthis, ittacitly refersto logic and generalexperience,-assuminghat the10ProfessorMalcolmSharp, my colleaguewho deals with contracts,expressedhisoffhandopinion hatthebadqualityof thepaintin the firstdrum shouldservethebuyeras an excusefor non-acceptance f a second drumof the "samepaint"-i.e., paintsoldunder hesame brand.

693


7/18

CALIFORNIA LAW REVIEWprinciples of reasoningare known to its judges and ministers, justas a vast multitudeof otherthingsareassumed as alreadysufficientlyknown to them.""Wigmore, agreeing with Thayer in much, questions the conclusionthat the law furnishes no test of relevancy,12 citing in opposition awell-known passage by Chief Justice Cushing of New Hampshire:

"... although undoubtedly the relevancy of testimony is origi-nally a matter of logic and common sense, still there are many in-stances in which the evidence of particularfacts as bearingon par-ticular issues has been so often the subject of discussionin courts oflaw, and so often ruled upon, that the united logic of a great manyjudgesand lawyersmay be said to furnish evidence of the sense com-mon to a great many individuals, and, therefore,the best evidence ofwhat may be properlycalled common-sense,and thus to acquiretheauthority of law. It is for this reason that the subject of the rele-vancy of testimony has become, to so great an extent, matter ofprecedent and authority, and that we may with entire proprietyspeak of its legal relevancy."13This question of the propriety of speaking of "legal relevancy",of the expediency of framing rules with the authority of law to govern

offers of evidence of particular facts as bearing on particular issues,is the central one in circumstantial proof. But since it is agreed onall sides that relevancy is at least originally a matter of logic andcommon sense rather than a matter of law, let us first see whitherlogic and common sense lead us, and by what means.

Dean Wigmore, recognizing this priority of logic, discusses theform of argument involved in the use of circumstantial evidence. Hisviews deserve extended quotation and criticism. After stating thatproof reduces to two great forms, inductive and deductive, and quot-ing Professor Sidgwick's description of the two, Wigmore states:

"A brief examinationwill show that in the offeringof evidenceinCourtthe form of argument s always inductive. Suppose,to proveachargeof murder,evidence is offered of the defendant'sfixed designto kill the deceased.The formof the argument s: 'A plannedto killB; therefore,A probablydid kill B.' It is clear that we have here nosemblanceof a syllogism.The form of argumentis exactly the samewhen we argue: 'Yesterday,Dec. 31, A slipped on the sidewalkandfell; therefore,the sidewalkwas probablycoated with ice'; or, 'To-11 THAYER, op. cit. supra note 1, at 265.12 1 WIGMORE,oc. cit. supra note 7.l State v. LaPage (1876) 57 N. H. 245, 288. Wigmore suggests that the differencebetween Thayer and Cushing is "one of nomenclature only", a very fundamental erroras should appear hereinafter.

694 [Vol.29


8/18

1941] RELEVANCY, PROBABILITY AND THE LAWday A, whowas bitten by a dog yesterday,died in convulsions; there-fore, the dog probablyhad hydrophobia.'So with all other legal evi-dentiary facts. We may argue: 'Last week the witness A had a quar-rel with the defendantB; therefore,A is probablybiassedagainst B';'A was found with a bloodyknife in B's house; therefore,A is prob-ably the murdererof B'; After B's injury at A's machinery,A re-pairedthe machinery; therefore,A probablyacknowledgedthat themachinerywas negligently defective'; 'A, an adult of sound mindand senses, and apparently impartial,was present at an affray be-tween B and C, and testifies that B struckfirst; therefore, t is prob-ably true that B did strike first.' In all these cases, we take a singleor isolated fact, and upon it base immediatelyan inference as to thepropositionin question."It may be replied, however,that in all the above instances, theargumentis implicitly based upon an understood law or generaliza-tion, and is thus capable of being expressedin the deductive or syl-logistic form.Thus, in the first instanceabove, is not the true form:'Men's fixed designs are probablycarriedout; A had a fixed designto kill B; therefore,A probably carried out his design and did killB'?"There are two answers to this. (1) It has just been seen thatevery inductiveargument s at least capableof beingtransmuted ntoand stated in the deductive form, by forcing into prominencetheimpliedlaw or generalizationon whichit rests moreor less obscurely.Thus it is nothing peculiarto litigious argumentthat this possibilityof turningit into deductive formexists here also. It is not a questionof what the formmight be-for all inductivemay be turnedinto de-ductive forms-, but of what it is, as actually employed; and it isactually put forwardin inductive form. (2) Even supposing thistransmutation to be a possibility, it would still be undesirable tomake the transmutationfor the purposeof testing probative value;because it would be useless. We should ultimately come to the samesituation as before.Thus, in one of the instances above: 'A repairedmachineryafter the accident; therefore,A was conscious of a negli-gent defect in it'; supposewe turn this into deductive form: 'Peoplewho make such repairsshow a consciousnessof negligence; A madesuch repairs; therefore,A was conscious of negligence.'We now havean argumentperfectly sounddeductively, i.e. if the premisesbe con-ceded. But it remainsfor the Courtto declare whetherit accepts themajor premise,and so the Court must now take it up for examina-tion, and the proponentof the evidenceappearsas its championandhis argumentbecomes: 'The fact that peoplemakesuch repairsindi-cates (shows, proves, probably shows, etc.) that they are consciousof negligence.'But here we comeagain,after all, to an inductive formof argument.The consciousnessof negligenceis to be inferred fromthe fact of repairs,-just as the presenceof electricity in the cloudswas inferred by Franklin from the shock through the kite-string,i.e. by a purely inductive form of reasoning.So with all other evi-

695


9/18

696 CALIFORNIA LAW REVIEW [Vol. 29dence when resolved into the deductive form; the transmutation isuseless, because the Court'sattention is merely transferredfromthesyllogismas a whole to the validity of the inferencecontainedin themajorpremise; which presents itself again in inductive form."For practical purposes, then, it is sufficient to treat the use oflitigious evidentiary facts as inductive in form."14Note, Wigmore does not deny that in every instance proof mustbe based upon a generalization connecting the evidentiary propositionwith the proposition to be proved.15 Conceding this, he argues thatthe generalization may as well be tacitly understood as expressed,that "the transmutation [from the inductive to the deductive form]

is useless, because the Court's attention is merely transferred fromthe syllogism as a whole to the validity of the inference contained inthe major premise".'1 Yet it is precisely in this transfer of attentionthat the value of the transmutation lies. The author's own examplesillustrate the point. In the case of the repaired machinery we aretold: " 'People who make such repairs [after an accident] show aconsciousness of negligence; A made such repairs; therefore, A wasconscious of neglience.' "'1 Before this deductive proof can be evalu-

14 1 WIGMORE,op. cit. s-upranote 7, at 416. Compare Starkie's equally surprisingobservation: "Circumstantial, or, as it is frequently termed, presumptive evidence, isany which is not direct and positive."An inference or conclusion from circumstantial or presumptive evidence may beeither the pure result of previous experience of the ordinary or necessary connection be-tween the known or admitted facts, and the fact inferred, or of reason exercised uponthe facts, or of both reason and experience conjointly. And hence such an inference orconclusion differs from a presumption, although the latter term has sometimes, yet notwith strict propriety, been used in the same extended sense. For a presumption in strict-ness is an inference as to the existence of one fact, from a knowledge of the existence ofsome other fact, made solely by virtue of previous experience of the ordinary connectionbetween the known and inferred facts, and independently of any process of reason inthe particular instance." STARKIE,EVIDENCE*478.15The argument thus far may be summarized as follows: Relevancy is formal rela-tion between two propositions. To determine the relevancy of an offered item of evidenceone must first discover to what proposition it is supposed to be relevant. This requiresanalysis of the express or tacit argument of counsel. Then, since evidence is admissibleonly if relevant to a material proposition, analysis of the pleadings and applicable sub-stantive law is required to determine whether the proposition ultimately sought to beproved is material. Having isolated the material proposition sought to be proved, westill must determine whether the evidentiary proposition is relevant to it-does tend toprove it. This tendency to prove can be demonstrated only in terms of some generalproposition, based most often on the practical experience of the judge and jurors as men,sometimes upon generalizations of science introduced into the trial to act as connectinglinks. If the evidentiary proposition itself is material, there is no problem of relevancy.16 1 WIGMORE,p. cit. supra note 7, at 417.17 Ibid.


10/18

1941] RELEVANCY, PROBABILITY AND THE LAWated, ambiguity must be eliminated from the major premise. By"people"shall we understand "somepeople"or "all people"?If theargumentis intended to read, "Somepeople who make such repairsshow consciousness of neglience; A made such repairs; therefore,Awas conscious of negligence,"it contains an obvious logical fallacy.If intended to read, "All people who make such repairs show con-sciousness of negligence; A made such repairs; therefore, A wasconscious of negligence," t is logically valid. However, few could befound to accept the premise that all persons who repair machineryafter an accident show consciousness of guilt; that is, that no singlecase could be found of one who, confident of his care in the past,neverthelessmaderepairsto guard againstrepetitionof an unforesee-able casualty or to preservefuture fools against the consequenceoftheir future folly. Here the result of transmutinga proposeddirectinference into deductive form is discoverythat it is invalid-at leastin the terms suggested.

The other proposed argument is equally interesting: "'Men'sfixeddesignsareprobablycarriedout; A had a fixeddesignto kill B;therefore,A probably did kill B."'18 Once one attempts to deal, ina quasi-syllogistic form, not with certainties but with probabilities,additionalopportunitiesfor fallacy are presented.Supposethat it isargued: "Most As are X, B is an A, thereforeB is probablyX"; or"Nine-tenths of all As are X, B is an A, therefore the chances arenine to one that B is X." Neither of these arguments s logically validexceptuponthe assumptionthat As may be treated as a uniformclasswith respectto the probabilityof their being X. This can be becausetherereallyis no way of sub-dividing heclass, findingmoreXs in onesub-class than in another,or because no subdivisioncan be made interms of availabledata. Supposethat nine-tenths of all people in theworld have dark eyes. If absolutely all one knew about B was thathe was a person, it would be an apparentnine-to-onechance that Bhad dark eyes. But if one knew B to be a Swede, the percentageofdark eyes in the total population of the world would no longer beimportant.One would want to know about the proportionof dark-eyed Swedes,which might differ from the ratio amonghumansgen-erally.'9 Similarlyin Wigmore'sexample.We knowthat we are inter-

18Ibid.19The form of the invalid argument would be as follows: Nine-tenths of men havedark eyes. B is a Swede. All Swedes are men. Therefore nine chances to one B has dark

eyes.

697


11/18

CALIFORNIA LAW REVIEWested in the probabilityof execution of a fixed design of a particularkind: to commit murder. There may be variation in the probabilityof execution of fixed designs on various subjects. As an initial criti-cism, therefore, the primary generalizationshould be "Men's fixeddesignsto kill areprobablycarriedout."In this formwe have a valid,quasi-syllogistic argument based upon the limited data available.2Still, is the premisesound?"Men'sfixeddesignsto kill areprobablycarriedout," as a majorpremisein this argument,must mean that they are carried out moreoften than not. Whilethe word "probable" anbe usedin othersenses,its meaninghere is clear. Hence one would conclude from the singledatum that A had a fixed design to kill B, no other evidence beingoffered,that morelikely than not A actually did kill B. But when thisargumentwas presentedto a groupof law students and teachers,onlyone was willing to accept the indicated conclusion. Several wouldaccept it if supported by adequate evidence that B had been inten-tionally killed by some one. Othersrefused to accept it without stillfurtherevidenceconnectingA with B's death,or at the very least evi-dence that B had no other enemies.Moreover, here was less hesitancyin acceptingthe argumentin its "inductive" form. Once the general-ization was made explicit, and particularly after discussion of themeaningof "probably"as thereused,doubts as to the proprietyof theinference arose or sharpened.The demonstration,however "valid",is no better than its majorpremise,and the more one considersthispremise the less reliable it looks. Certainly a permitted inferenceshould rest uponsomemoreeasily acceptable aw.Of course, it does not followthat a proposed nferenceis improperbecause it can be shown not to follow on the basis of one possiblegeneralization,or becauseanother-which by the rules of logic wouldvalidate the inference-is unacceptable.There may be a thirdlaw, asyet unexpressed,which would justify the inferenceand at the sametime be commonly acceptedas true. And it may be very importanttofindthe valid and accuratelink, since the formof the link will controlthe formof the conclusion.Persons who are unwilling to agree that men's fixed designs (atleast in case of murder) are "probably"carriedout-or, even con-

20Probability is not an actual state. Nothing really is probable. It is true or false.Probability is a matter of appearance. Apparently probability is always relative to thedata available at the time judgment is exercised. If all possible data were available weshould be dealing not with probability in an ordinary sense but with the approximationof certainty.

698 [Vol.29


12/18

RELEVANCY, PROBABILITY AND THE LAWceding the fact of murder,that proof of A's fixed design to kill Bestablishes A, more likely than not, as B's killer-still agree thatsomehow this bit of evidence does have some tendency to indicateA's guilt. What formof generalstatementcan reconcilethese views?Perhaps something like this: "Men having such a fixed design aremore likely to kill than are men not having such a fixed design."Those who contend that even fixed designs to kill are more oftenabandonedor thwartedthan carriedout can and doubtless will stillconcedethat enoughsuch designsare carriedto executionso that thepercentageof murderers s higheramong personsentertainingsuch afixeddesign than amongthe generalpublic. Obviouslythis proposedgeneralizationdoes not lead us from A's fixed design to kill B to theconclusionthat A probablydid kill B. There is nothingdisturbinginthis. This conclusion simply does not follow from the evidence ofdesign. The errorwas in the original"direct induction".In fact, nouseful conclusion about A's guilt can be drawnfromdesign or intentalone. On the basis of an acceptable generalizationwe are able onlyto place A in a class of persons in which the incidence of murderisgreaterthan amongthe generalpublic. We cannot now say that A isprobably guilty, but we can say that the apparent probabilityof hisguilt is nowgreaterthan beforethe evidenceof designwas received.2This is logical relevancy-the only logical relevancywe can expect indealingwith practicalaffairs where strict demonstration s neverpos-sible.2 The advantage of the transmutationinto deductive (thoughnot strictly syllogistic) formis that we know to what degreeof proofwe have attained, and do not overstate our results.23Before leaving the subject of induction and deduction, consider

21 Or, the apparent probability of his innocence is now less. The two statementsmean the same.22 It should be unnecessary to point out the fundamental distinction between rele-vancy of evidence and the degree of probability necessary to establish a prima facie case.Almost every careful writer has discussed it. See, for example, 1 WIGMORE,op. cit. supranote 7, ?? 28, 29. If the sum total of evidence received falls short of satisfying the burdenof persuasion, the jury is there for the express purpose of finding against the party whomust bear that burden. If the sum total of evidence received is so slight that the jury asreasonable men could not be persuaded, the trial judge may enter a non-suit or direct averdict. There is no reason for worrying about such issues while passing upon an offer ofevidence. The one exception might be an item of evidence which, although logically rele-vant in a very remote sense, is so slight that in the opinion of the trial judge it could notsway an even balance.

23 In some cases we may be able to discover no acceptable generalization to connectthe offered evidence to any material proposition. Such cases will be very rare, unless theyinvolve errors of pleading or substantive law.

19411 699


13/18

CALIFORNIA LAW REVIEWone more example.Wigmoreasserts: "There is just as much proba-tive value in the argument'A is quarrelsome,thereforehe probablycommitted this assault,' as in the argument, 'B is peaceable, there-fore he probablydid not commit the assault....' "' So it may ap-pear when the two argumentsare stated as direct inductions. Buttransmutationof the two argumentsshows a differentresult. In thesecond case we say: "Peaceable men are not likely to commit as-saults. B is a peaceableman. Thereforehe was not likely to commit(this or any) assault." In the first case the most we can say is:"Quarrelsomemen are more likely than others to commit assaults.A is quarrelsome.Therefore he is more likely than others to com-mit assaults." This may be equally good if there is no doubt of anaffraybetweenA and B, one of them surelybeing the aggressor.Butsuppose that we know someone assaulted C and are attempting todecide whether A or B or some third person was the culprit. Weknow that B, the peaceable man, was unlikely to commit any as-sault, hence unlikely to have committed the one in issue. All weknow of A, the quarrelsomefellow, is that he has a generalpredi-lection for trouble.While it is somehelp, it does not show an affirma-tive probabilityof his guilt in any particularcase. There is a definitedifferencein the probativevalue of the two offers of proof.Nevertheless, both offers of proof are logically relevant; the ap-parent probability of guilt in each case would be changed, thoughnot in the same degree, by the evidence of character.Why then isthe evidence rejected in the weaker case? A possible answer wouldbe simply because it is weaker-too weak. Although logically rele-vant, it falls short of the minimumrequirementof legal relevancy.25Such an answer merely raises a furtherquery. Why should there bea standardof "legal relevancy"more strict than, or in any respectdifferentfrom, the standards of logic? Why exclude any data whichif admitted wouldchangethe apparentprobabilitiesand hence serve,even to a slight degree, to aid the search for truth? Justice Holmessuggested one answer, it is "a concessionto the shortnessof life"20

24 1 WIGMORE,p. cit. supra note 7, at 454.25Bouvier, after defining "relevancy in a logical sense" goes on to say "Legal rele-vancy requires a higher standard of evidentiary force. It includes logical relevancy anddemands a close connection between the fact to be proved and the fact offered to proveit. The fact, however, that it is logically relevant does not insure admissibility; it mustalso be legally relevant." 3 BOUVIER, DICTIONARYRawle's 3d ed. 1914). BLACK,LAWDIcrIONARY3d ed. 1933), is to the same effect.

26Reeve v. Dennett (1887) 145 Mass. 23, 28, 11 N. E. 938, 944.

700 [Vol. 29


14/18

RELEVANCY, PROBABILITY AND THE LAW-and perhapsto the shortness of purse of harassedlitigants. If anyand all evidence may be admissible which-in terms of some com-monly accepted generalization about human conduct or naturalevents-would operate to any extent to alter the apparent prob-ability of some material proposition,the field of judicial inquiry inmost cases would be almost unlimited. Trials could come to an endonly by the exhaustion of lawyers' ingenuity or clients' money, andthe trial judge or jury might be overwhelmedand bewilderedby themultiplicity of collateral issues. Such a rule would result in the ap-parent justice and the practical injustice characteristicof EnglishChancerypracticea centuryand a half ago. Hence the requirement,in many cases, of somethingmore than bare logical relevancy.In addition to the simple factors of time, confusion and expense,other practical considerationsmay justify the exclusion of relevantevidence of but slight probativeweight.' As noted above,evidenceisirrelevantin the clearest sense if its tendencyis to provesomepropo-sition not properly provable in the action. Sometimesthe same evi-dence may be remotely relevant to a material proposition and di-rectly relevant to an immaterialone; in such a case its exclusioncould be justified lest the jury be misled into deciding the case onthe immaterialissue. Or evidence might be excluded because underthe issues as framedin the pleadingssuch evidencewouldunreason-ably surprise the opponent.28

Probably the greatestpart of judicial rulings excludingevidenceas irrelevantgo primarilyon the firstprinciple-that the evidenceisrelevant,but its probativevalue is so slight as not to justify the timeand expense involved in receiving it and the confusion of issueswhich might result in the mind of the trier of fact. Does this soundpractical policy justify creation of the concept of "legal relevancy"-higher and more strict than logical relevancy, to which offers ofproof must be referred?On the contrary,the concept can only be anuisance. In the first place, it defies definition.2 No statement of a

27 All of these factorsare listedas permissible rounds or the exclusionof evidencein the proposedCode,op. cit. supranote 3, Rule 403, whichRule is expresslymadeaqualification f most otherrules,but is then followedby severalspecificand somewhatconfusing ules on proofof character, abit,etc.28A similarpolicyis expressedby pleadingrules whichrequire he affirmative lle-gation of defenses ikelyto surprise he opposingparty,whetheror not the defense saffirmative s a matterof analysis.See,e.g.,ILLrtoisCvILPRACTICECT1933)?43(4).2 Seesupranote 25. Sucha definition s sufficiento raisedoubts n everycaseandto satisfythem in none.

19411 701


15/18

CALIFORNIA LAW REVIEWstandard of probative value higher than logical relevancy can bemadepreciseenoughfor use withoutexcludingmuchevidencewhichis used daily, without argumentand to good effect. And there is adefinite reason for this. When a judge must decide whether a par-ticular item of evidence, logically relevant, is of sufficient force tojustify the time and expense necessaryto establish it, he should notconfine his attention to the effect of the offeredevidence alone. Heshould consider how difficult it will be to establish the evidentiaryfact-whether there is any real contest about it and whether con-fusing side issues must be explored. He should consider how theofferedevidence may fit in with other evidence. It may form a smallbut useful part of a patternof proof. It may stand alone, lending itsnegligibleaid to unconnectedlines of proof. It may be merelycumu-lative, so that the trouble of establishing it will result in little ofpracticalvalue. He may even want to considerthe importanceto theparties of the issues being tried, in the light of mounting trial ex-pense. Such considerationscannot usefully be reduced to a simpleformula for relevancy of particularitems of evidence.30The methodsuggestedby Chief Justice Cushing,31hat of build-ing up a body of rulings on the bearing of particular facts uponparticular issues, is no better-for the same reason. The ruling ineach case will seldom if ever be that the offered fact has no bearingupon the issue. It will be that it has no sufficientbearing to justifyits use in view of all the circumstancesof the case in which it isoffered.Such a precedent, rightly understood,is of no value save inanothercase substantially identical in all of its particulars.Treatedmore broadly, its tendency will be to mislead subsequent judges.

Take, as an example,State v. LaPage32where Chief Justice Cush-ing enunciatedhis theories on legal relevancy. The state, in a prose-cution for murderof one Josie L. in which there was some evidenceof murderduringan attemptedrape,was permittedto show that thedefendant had raped one Julienne R. The evidence was offered not

30As good an attempt as any was that of Justice Cooley in Stewart v. People(1871) 23 Mich. 63, 75: "The proper test for the admissibility of evidence ought to be,we think, whether it has a tendency to affect belief in the mind of a reasonably cautiousperson, who should receive and weigh it with judicial fairness." This is practically adefinition of logical relevancy applied to important issues depending on indeterminateprobabilities.

31 Supra note 13.32Ibid.

702 [Vol. 29


16/18

RELEVANCY, PROBABILITY AND THE LAWto show that the defendant had killed Josie, but to show that if hehad done so it was in the course of an attemptedrape and thereforefirst degree murder under the New Hampshirestatute. A convictionwas reversed, the court holding that evidence of another crime, notrelevantsave as showingthe characteror dispositionof the accused,could not be presentedby the prosecutionwhen the accused had not"put his characterin issue". The reasonsfor this rulingare clear onprinciple and from the opinion itself. First, the offered evidence,while it had some probativevalue, was not particularlycogent. Sec-ond, by attacking the characterof the accused, it raised the dangerthat he might be "overwhelmedby prejudice, instead of being con-victed on that affirmative evidence which the law of this countryrequires".33 hird, by using specificinstances of misconductto dem-onstrate bad character,it confronted the accused with a dangerousissue which he might have been entirely unpreparedto meet. Theseare substantial argumentsfor exclusion,which are in no way aidedby saying that the defendant's bad character is not "legally rele-vant", or that specific instances of misconductare not "legally rele-vant" to prove bad character.On the contrary,there is real dangerthat talk of legal relevancyin such cases may mislead other judges. Thus in a prosecutionformurder the character of the decedent as quarrelsomeand violentshould be admissible in aid of a plea of self-defense. Most courtsadmit it,34but those who do not often seem confused by the inad-missibility of the violent character of the accused. If the ruling ex-cluding the bad character of criminal defendants were always ex-pressly made on the ground that such evidence is relevant but in-admissible because unduly prejudicial, without reference to "legalirrelevancy",such confusion would be unlikely.

Similarly the argumentof undue surprise is less likely to leadcourts into error if stated as such. Specificimmoral acts clearly arenot irrelevantas evidence of lack of the correspondingvirtue. If notactually conclusive they are certainly much more cogent than repu-tation. If they are really excludedbecause of the dangerof surprise,it would be better to say so. Then courts might in time discover inpre-trial practice a method of obviating unfair surprise in the useof the specific instance whenever its value is sufficiently great tojustify the time and expense of trying a collateral issue.33 Regina v. Rowton (1865) Leigh & C. 520, 540, 169 Eng. Rep. 1497, 1506.34 1 WIGMORE,p. cit. supra note 7, ? 63.

19411 703


17/18

CALIFORNIA LAW REVIEWSuch considerations, whether taken singly or together, affordmost unsuitable materials for the construction of a body of case

law. Like the simpler problem of time and expense, all involve abalance of competingconsiderations-the value of the evidence, theimportanceof the point sought to be established,the availability ofother evidence, the precise degree of inconvenienceor prejudice-never likely to be duplicated in different cases. "The evidence ofparticularfacts as bearing on particularissues" does not repeat itspattern closely enough to permit its petrifaction into rules having"the authority of law" without over simplification.The history ofevidence has been in the developmentof sound principles into arbi-trary and unworkablerubrics,a developmentnot to be encouragedin case law or by code.The remedylies in Thayer's advice in holding to logic and gen-eral experience,assumingthat the principlesof reasoningare knownto the judges and ministers of the law. In this analysis evidence isrelevant if it can be demonstrated,in terms of some generalizationacceptable to the court, to alter the apparent probability of anymaterialpropositionin the case. The generalizationmay rest eitherupon the general experienceof the court or upon expert testimony,but it should be susceptibleof expression.Becauseof the wide rangeof relevantevidenceundersuch a test,the trial judge must have discretionto exclude it, and may well befurnished with general criteria to aid, but not to control, the exer-cise of this discretion.Although it is rather brief, Rule 403 of theproposedCode is a step in this direction.Attempts at furtherspeci-fication, either by code or by decision, should then be discouraged.Limiting the review of trial court rulings on circumstantialproofwill do muchto check the proliferationof precedent.Virtuallyneverwouldan appellatecourtbe justifiedin reversinga trial judge for theadmissionof remotebut non-prejudicialevidence.Time thus wastedcannot be recoveredby requiringa second trial. Where remote evi-dence was excluded,an upper court should refuse to reverse unlessit appears that a contrary ruling would actually have altered theresult. To the extent that this rule allows somewhat more controlthan the one on erroneousadmission,its tendencywould be to liber-alize rulings on evidence where the only objection is remoteness.Finally no judgment should be reversed because evidence wasprejudicial,or caught the opponentby surprise,unless the objectionwas made on this express ground. If problemsof surpriseor preju-

704 [Vol. 29


18/18

1941] RELEVANCY, PROBABILITY AND THE LAW 705dice were necessarily designated as such and not lumped under theheading of irrelevancy, the bench and bar might learn, instead ofregardingthem primarilyas ammunition of appeal, to handle themat the trial stage by instructions to the jury, by continuances,andabove all by discussion of the problems of proof at pre-trial con-ferences. The dominationof the appellate court, the printed reportand the misleading precedentmight be ended, the largely unrelatedprinciples making up the concept of "legal relevancy"be disentan-gled, and that ambiguousphrase returnedto the grave whereinthegreat Professor Thayer laid it almost fifty years ago.

Documents

Probability and the Law