Journal of Criminal Law and CriminologyVolume 71Issue 4 Winter Article 3

Winter 1980

Harmless Error: Constitutional Sneak ThiefSteven H. Goldberg

Follow this and additional works at: https://scholarlycommons.law.northwestern.edu/jclc

Part of the Criminal Law Commons, Criminology Commons, and the Criminology and CriminalJustice Commons

This Criminal Law is brought to you for free and open access by Northwestern University School of Law Scholarly Commons. It has been accepted forinclusion in Journal of Criminal Law and Criminology by an authorized editor of Northwestern University School of Law Scholarly Commons.

Recommended CitationSteven H. Goldberg, Harmless Error: Constitutional Sneak Thief, 71 J. Crim. L. & Criminology 421 (1980)

0091-4169/80/7104-0421$02.00/0THE JOURNAL OF CRIMINAL LAW & CRIMINOLOGY

Copyright © 1980 by Northwestern University School of Law

HARMLESS ERROR: CONSTITUTIONAL SNEAK THIEF

STEVEN H. GOLDBERG*

"Harmless constitutional error" is among themost insidious of legal doctrines. Since its promul-gation by the United States Supreme Court in

Chapman v. Californial it has determined as manycriminal appeals as have some of the more well-known and hotly debated decisions of the 1960s. 2

Despite the frequency of its use in determiningcriminal appeals-possibly as high as ten percent

of all criminal appeals during the last thirteenyears--it has received comparatively little critical

* Assistant Professor, University of Arkansas at Little

Rock School of Law; J.D., University of Minnesota, 1968;B.A., Northwestern University, 1963.

1386 U.S. 18 (1967).2 It is probably impossible to determine exactly how

many cases have been "determined" by a prior precedent.Shepard's United States Citations lists in excess of 6000citations to Chapman v. California. While some of thosecitations probably represent cases which found the error"harmful," Chapman is almost always cited as authorityfor an appellate court's finding of "harmlessness." Arough comparison with the page measure of the citationsto Mapp v. Ohio, 367 U.S. 643 (1961), and Miranda v.Arizona, 384 U.S. 436 (1966), two of the major decisionsof the Warren Court, supports the proposition that Chap-man has determined as many cases as almost any prece-dent from the decade. While both Miranda and Mapp arecited more often, they are more often distinguished orexplained and frequently appear numerous times in thesame case.

3 The empirical research to establish the exact rela-tionship between harmless constitutional error determi-nations and all criminal appeals is probably not worththe effort. The available data and a couple of assumptionsmake a fair argument for the likelihood of the 10 percentestimate. While the Shepard's citations undoubtedly in-clude some cases of harmful error and some multiplecitations in the same case, a number of cases are deter-mined by a finding of harmless constitutional error with-out any mention of Chapman. Harrington v. California,395 U.S. 250 (1969), provided a less rigorous test thanChapman and is often cited in preference to Chapman.Similarly, many state courts cite the state authority forharmless constitutional error rather than Chapman. It isnot unreasonable to assume that the cases which do notcite Chapman for the harmless error proposition are atleast as numerous as those that do and do not find theerror harmless. Something in excess of two-thirds of theShepard's citations to Chapman are in state cases.

The available data for harmless error as a percentageof appellate decisions is found in Note, Harmful Use ofHarmless Error in Criminal Cases, 64 CORNELL L. REv. 538(1979) (hereinafter Harmful Use). The Note presents acomparison of all federal cases mentioning harmless errorto all cases in the federal circuits from 1960 through 1978.

attention.4 The reason for the inattention? It's a

sneak thief. Its appearance does not raise appre-hension, and its application does not leave concen-

trated areas of obvious constitutional damage. Thedoctrine does not aim at any closely guarded right.

It poses no consistent doctrinal challenge to impor-tant judicial determinations; nor does it consist-

ently affect any police practice. Further, it lookslike the helpful, familiar doctrine of harmless error.The purpose of this article is to demonstrate that

The data do not discriminate between harmless error andharmless constitutional error. The markedly increasedpercentage, and the even more dramatic increase inabsolute number of harmless error cases following theChapman opinion, when combined with the observationthat there is a rough approximation between the numberof federal harmless error cases presented in Harmful Use,supra, and the federal citations in Shepard's for the sameperiod, lead to the conclusion that the number of harm-less error cases which are not harmless constitutionalerror is statistically insignificant.

During the last decade, civil cases outnumbered crim-inal cases in the federal circuits approximately four-to-one. See DEPARTMENT OF COMMERCE, STATISTICAL AB-STRACt OF THE UNITED STATES (1979). In Harmful Use,supra, the author estimates that 2.5 percent of all appellatecases in the federal system are harmless error cases. If thecivil-to-criminal ratio in the states is comparable to thatof the federal system, then approximately 10 percent ofall criminal appellate cases throughout the country aredetermined by a finding of harmless constitutional error.Granting that the figure represents the roughest of ap-proximations, the harmless constitutional error doctrineis apparently a significant factor, at least by volume, incriminal appellate decisionmaking.

4 When compared to the other major decisions of the1960s, the harmless error case was ignored. The mostcomplete discussion of harmless error is R. TRAYNOR, THE

RIDDLE OF HARMLESS ERROR (1970). Field, Assessing theHarmlessness of Federal Constitutional Error-A Process in Needof a Rationale, 125 U. PA. L. REv. 15 (1977); Mause,Harmless Constitutional Error: The Implications of Chapman v.California, 53 MIss. L. REV. 519 (1969); and Saltzburg,The Harm of Harmless Error, 59 VA. L. REv. 988 (1973),each deal with a particular aspect of the harmless consti-tutional error problem. A handful of student notes andcomments, the most complete of which is Note, HarmlessConstitutional Error, 20 STAN. L. REv. 83 (1967), com-mented on the Chapman decision. While many of thecommentators presented a case for excluding one kind oferror or another from the reach of the harmless consti-tutional error doctrine, none criticized the doctrine asbeing simply wrong.

Vol. 71, No. 4Pnlnted in U.S.A.

STEVEN H. GOLDBERG

the doctrine of harmless constitutional error de-

stroys important constitutional and institutionalvalues and therefore should be discarded.

I. HARMLESS ERROR/HARMLESS CONSTITUTIONAL

ERROR

The lack of apparent concern for the doctrine ofharmless constitutional error is attributable, atleast in part, to the assumption that harmless con-

stitutional error is simply another variety of harm-less error. In fashioning a "harmless-constitutional-error rule"5 for the seven-member Chapman major-ity, Justice Hugo Black made a special attempt to

tie his new rule to the harmless error statutes extant

in the various jurisdictions: "All 50 States haveharmless-error statutes or rules, and the United

States long ago through its Congress establishedfor its courts the rule that judgments shall not bereversed for 'errors or defects which do not affect

the substantial rights of the parties.' ,6

Harmless constitutional error is different fromharmless error generally, and that difference begins

with the statutes and rules in the fifty states. Thevarious harmless error statutes state, in one fashion

or another, that a trial result shall not be reversedon appeal unless the trial included a "substantial

wrong or miscarriage.",7 The language and the

concept originated in England. The method andreason for its eventual transplantation to theUnited States bears upon whether harmless erroris a persuasive precedent for the harmless consti-

tutional error doctrine.Commentators generally believe the 1835 case of

Crease v. Barrett created the Exchequer Rule whichstates that prejudice presumptively attends everytrial error.9 The rule resulted in such overcrowding

in the English courts that litigation seemed to

survive until the parties expired. Faced with moreretrials than new trials, the English created a harm-less error rule for civil litigation which prohibited

reversal absent substantial wrong.The American courts adopted the Exchequer

" 386 U.S. at 22. The phrase is Justice Black's owndescription of what Justice Stewart considered to be a"break with settled precedent." Id. at 22 (quoting id. at45 (Powell, J., dissenting)).

6Id. at 22 (quoting 28 U.S.C. 2111 (1966)).7 See 36 & 37 Vict., c. 66, rule 6, sched. 1. (1873)." I C.M. & R. 919, 149 Eng. Rep. 1353 (Ex. 1835).9 Wigmore first blamed Crease for the rule, and most

commentators have accepted that genealogy. R. TRAY-

NOR, supra note 4, at 4, argues persuasively that theExchequer Rule was not invented by Baron Parke inCrease, but rather by the judges who misread the prece-dent in applying Crease to the case of the moment.

Rule and developed the same backlog and delay

that plagued the English. Unlike the English, how-ever, the American courts did not change the ruleand even in the early twentieth century were stillleaving no error unremedied, no matter how incon-sequential. Commentators labeled the courts "im-pregnable citadels of technicality,"'1° and by allaccounts the label was warranted. Cases were oftentried more than once-and once, five times.i Con-victions were overturned for matters as inconse-quential as the omission of the word "the" beforethe words "peace and dignity" in an indictment. 12

One particularly glaring example of delay involveda widow who, twenty-three years after filing suitfor the proceeds of her husband's life insurance,appeared before the Supreme Court for the secondtime.' 3 Many lawyers placed error in the record asa hedge against losing the verdict. 4 The situationin the courts became intolerable to many membersof the organized bar, some judges, and a numberof legal scholars. They formed a loose coalition topress for remedial legislation.is The reform move-ment resulted in "harmless error" legislation invirtually everyjurisdiction.5 6 In the words ofJustice

10 Kavanaugh, Improvement of Administration of Criminal

Justice by Exercise of Judicial Power, 11 A.B.A.J. 217, 222(1925).

" Pressley v. Bloomington and Normal Ry. & LightCo., 271 Ill. 622, ill N.E. 511 (1916), was retried fourtimes without a substantial error.

12State v. Campbell, 210 Mo. 202, 109 S.W. 706(1908).

13 Connecticut Mutual Life Ins. v. Hillmori, 145 U.S.285 (1892), 188 U.S. 208 (1903), might have become anational tragedy had the "widow" not been very young-and thereby likely to outlive the lawsuit to spend herwinnings-and had there been no suspicion that herhusband was not really a "corpse" and she, therefore, notreally a "widow."

14 There was considerable pressure to create a federalharmless error statute with application to civil cases only.See S. REP. No. 1066, 62nd Cong., 2d Sess. (1911-12).Criminal cases were nonetheless included because, ac-cording to Justice Rutledge in Kotteakos v. UnitedStates, 328 U.S. 750, 759 (1946), the "criminal trialbecame a game for sowing reversible error in the record,only to have repeated the same matching of wits when anew trial had been thus obtained."

"5The organized bar spearheaded the coalition, aidedby its powerful, if not succinctly named, Special Com-mittee to Suggest Remedies and Formulate ProposedLaws to Prevent Delay and Unnecessary Cost in Litiga-tion. Pound, Taft, Wigmore, Hadley, and Frankfurterwere only some of the legal community leaders involvedin the coalition.

16 Some of the harmless error rules, for instance theCalifornia constitutional harmless error rule, antedatedthe reform movement. Most of the statutes and rules,however, were passed during the quarter century in whichthe coalition was actively involved in lobbying.

[Vol. 71

HARMLESS ERROR" CONSTITUTIONAL SNEAK THIEF

Frankfurter the purpose and limits of the harmlesserror legislation were: "to prevent matters con-cerned with the mere etiquette of trials and withthe formalities and minutiae of procedure fromtouching the merits of a verdict."' 7

With this background, the issue of harmlessconstitutional error came to the Supreme Court in1963. Until Fahy v. Connecticut,'0 no court had sug-gested that a federal constitutional error might beharmless. The Court had never given any seriousconsideration to the question, and no constitutionalerror had gone unremedied by reversal.'" Fahycame to the Court in the backwash of Mapp v.Ohio20 as the states were beginning to struggle withthe proposition that the fourth amendment limi-tations upon search and seizure, and the exclusion-ary rule remedy for violation of those limitations,were fully applicable to the states. The statecharged Fahy and a friend with painting swastikason a synagogue in violation of its law againstwillful injury to public buildings. At Fahy's trial,the state introduced the paint and brush used forthe artwork. On appeal, the Connecticut court heldthat the search for and seizure of the paint andbrush from the Fahy garage violated the fourthamendment and that Mapp required their exclusionfrom evidence.2' The court refused, however, toreverse the conviction, holding that the error in theadmission of the paint and brush had not "mater-ially injured the appellant" under Connecticut'sharmless error rule.'s

The Supreme Court might have pursued severalalternatives in Fahy. It could have decided thatfederal constitutional error never could be harm-

17 308 U.S. 287, 294 (1939).18 375 U.S. 85 (1963).19 See, e.g., Gibbs, Prejudicial Error: Admissions and Exclu-

sions of Evidence in the Federal Courts, 3 VIL. L. REv. 48, 67(1957).Justice Harlan, dissenting in Chapman, cited Motesv. United States, 178 U.S. 458 (1900), in response toJustice Stewart's claim that constitutional error had al-ways led to reversal. 386 U.S. at 50. Saltzburg, supra note4, at 1001, suggests that Motes may not have been viewedas anything more than a waiver of constitutional rightsthrough Motes' admission of guilt, which amounted to aguilty plea. Regardless of Motes' "true" meaning, thematter of harmless constitutional error was never raisedor argued in the Supreme Court before Fahy. JusticeRutledge, in Kotteakos v. United States, 328 U.S. at 764-65, said in dictum that only constitutional norms andspecific commands of Congress could insulate error fromthe federal harmless error statute.20 367 U.S. 643 (1961).21 Connecticut v. Fahy, 149 Conn. 577, 586, 183 A.2d

256, 261 (1962).22 Id. at 588, 183 A.2d at 262; CONN. GEN. STAT. § 52-

265 (1958).

less. It could have decided that courts should judgefederal constitutional error by a federal standardof harmlessness. It might have separated the exclu-sionary rule from the fourth amendment and de-termined that the admission of the evidence wasnot constitutional error. The Court chose none ofthese alternatives. Instead the Court held that Fahywas entitled to a new trial because the Connecticutcourt wrongly believed that the admission of theevidence would not prejudice Fahy.23 Chief JusticeWarren, writing for the majority, failed to clarifythe basis of the Court's power to instruct a state onthe interpretation of its own laws.2' More impor-tantly, the opinion is not clear on whether Con-necticut read its facts right and its test wrong or itstest right and its facts wrong.25 The single clearproposition to be gleaned from Fahy is that, hadthe Connecticut court done its job correctly, itmight have found the federal constitutional errorto be harmless.

Chapman v. California,0 the "harmless error" case,came four years after Fahy. Significantly, Chapmandid not involve a fourth amendment violation.Justices Stewart and White, who had joined JusticeHarlan's dissent in Fahy, did not join him in Chap-man. Insofar as Justice Harlan's position was notsignificantly different on the relationship betweenharmless error and constitutional error, it is fair toconclude that the fourth amendment exclusionaryrule presents special problems for harmless error.2 7

2 "From the foregoing it clearly appears that theerroneous admission of this illegally obtained evidencewas prejudicial to petitioner and hence it cannot becalled harmless error." 375 U.S. at 91-92.

24Justice Harlan in dissent questioned the Court'spower to' instruct Connecticut upon the proper interpre-tation of its own harmless error statute. "Evidentiaryquestions of this sort are not a proper part of this Court'sbusiness, particularly in cases coming here from statecourts over which this Court possesses no supervisorypower." Fahy v. Connecticut, 375 U.S. at 92 (Harlan, J.,dissenting).

25 The majority opinion goes out of its way to discussthe facts. However, as the dissent points out, reinterpret-ing the facts is not the task of the United States SupremeCourt. Although Connecticut's statute is potentially asrigorous for harmless error as any test the Court mightpromulgate, the opinion states a harmless error test dif-ferent from that which Connecticut applied under itsown statute. The test which Warren stated in framingthe "factual" issue placed the burden on the state to showno "reasonable possibility" that the error contributed tothe verdict. It is unlikely that Connecticut read its harm-less error test similarly.

26 386 U.S. 18 (1967).2 The exclusionary rule presents special problems in

a number of contexts. A familiar contention is that mostof the procedural cases, see, e.g., Linkletter v. Walker, 381

STEVEN H. GOLDBERG

Chapman involved California's constitutional rulewhich allowed the prosecutor to argue to the jurythe defendant's failure to testify.28 Chapman anda codefendant were charged with a number ofcrimes, including murder. They chose not to testifyat their trial, and the prosecutor made a vigorousargument based upon their failure to explain andcontradict the state's evidence. After their trial, butbefore their appeal, the United States SupremeCourt decided Griffin v. Califomia.29 In Griffin, theCourt held that the California comment rule un-constitutionally eroded the fifth amendment priv-ilege against self-incrimination by making its asser-tion costly. The California Supreme Court recog-nized the applicability of Griffin to Chapman'scase, but held that under California's test for harm-less error the comment to the jury was harmless.3

0

The Chapman opinion is, in its own fashion, asstrange as Fahy. Justice Harlan, who did not un-derstand the source of the Fahy Court's power toinstruct Connecticut on its own law, had similardifficulty with the source of the Chapman Court'spower to create a federal harmless error standard,particularly, given Justice Black's observation thatCongress might impose a different standard. 31 Heconcluded that the majority, in a "startling consti-tutional development" had assumed "a generalsupervisory power over the trial of federal consti-tutional issues in state courts." 32 Former ChiefJustice Traynor of the California Supreme Courtread Justice Black's opinion quite differently, con-cluding that the Chapman rule was not constitution-

U.S. 618 (1965), and Stone v. Powell, 428 U.S. 465(1976), represent reactions to the exclusionary principlerather than antipathy to the right involved. Chapman,although not an exclusionary rule case, was decided withthe exclusionary rule in the very near background. Fur-ther, the exclusionary rule significantly affected the de-cision to the extent that it tempered Stewart's otherwiseuncompromising position that constitutional error wasdifferent as a class for purposes of harmless error.

' CAL. CONsT. art. I, § 13.29380 U.S. 609 (1965).3o People v. Teale, 63 Cal. 2d 178, 404 P:2d 209, 45

Cal. Rptr. 729 (1965).31 386 U.S. at 45 (Harlan, J., dissenting). The limits of

federal power-and certainly court power-is the centraltheme ofJustice Harlan's dissent. He makes the argumentmore completely than he did in Fahy. His dissent, andthe failure of Justices Stewart and White to join it, is thebest argument for two propositions: (1) the exclusionaryrule alone accounted for the Court's avoidance of theharmless error issue in Fahy, and (2) the Chapman deter-mination that federal constitutional error requires a dif-ferent and more rigorous harmless error test is a "consti-tutional judgment."

2Id. at 45, 46.

ally based but rather was an interpretation of the

federal harmless error statute. 3sMuch of the difficulty with Chapman is the result

of its posture in the Supreme Court, the positionsof the parties, and the Court's response to those

positions.34 The California Supreme Court appar-ently had concluded that the evidence of guiltpresented to the jury was so "overwhelming" that

the error was harmless.3s The state contended thatits harmless error rule represented the proper test.

The defendant contended that no constitutional

error could ever be harmless.36 The majority opin-ion rejected both contentions and fashioned a fed-

eral harmless error standard. In so doing, JusticeBlack held that federal constitutional error was

sufficiently different from common error to requirea different standard,37 but not sufficiently differentto prohibit any standard at all.38 For the latter

33 R. TRAYNOR, supra note 4, at 38-42.4 In the most technical sense, the language in Chapman

concerning the nature of the harmless constitutional errormay be considered dictum. Chapman is the authority usedby many appellate courts to find harmless constitutionalerror, but Chapman did not find such an error. TheCalifornia court found the error to be harmless; theUnited States Supreme Court found only that Califor-nia's use of its own test for harmlessness was wrong.

3 Whether Chapman disapproved the California test orthe way the California courts were interpreting that testis unclear. The issue of whether there is a differencebetween the "overwhelming evidence" and the "effect ofthe evidence" tests, and whether that difference matters,is born in Chapman: "The California-constitutional ruleemphasizes 'a miscarriage of justice,' but the Californiacourts have neutralized this to some extent by emphasis,and perhaps overemphasis, upon the court's view of'overwhelming evidence.' We prefer the approach of thisCourt in deciding what was harmless error in our recentcase of Fahy v. Connecticut." 386 U.S. at 23.

"6 The defense maintained an alternative position thatif all constitutional error did not require automatic re-versal, the error in their particular case was prejudicial.That position eventually prevailed. For the purpose ofthis article, the first position, automatic reversal, is ofinterest.

37 With faithfulness to the constitutional union ofthe States, we cannot leave to the States the for-mulation of the authoritative laws, rules, and rem-edies designed to protect people from infractions bythe States of federally guar.nteed [constitutional]rights.... In the absence of appropriate congres-sional action, it is our responsibility to protect [fed-eral constitutional rights] by fashioning the neces-sary rule.

386 U.S. at 21.38 All of these rules, state or federal, serve a veryuseful purpose insofar as they block setting asideconvictions for small errors or defects that havelittle, if any, likelihood of having changed the resultof the trial. We conclude that there may be someconstitutional errors which in the setting of a par-ticular case are so unimportant and insignificant

[Vol. 71

HARMLESS ERROR: CONSTITUTIONAL SNEAK THIEF

proposition, he relied upon the harmless error stat-utes of the fifty states, observing that they failed todistinguish "between federal constitutional errorsand errors of state law or federal statutes andrules."39 These were the same harmless error stat-utes which he found inadequately designed to pro-tect federally guaranteed rights.40 Commentatorshave observed that Chapman might be read to rejectthe method of measure-"overwhelming evi-dence"-the California court used in applying itsharmless error statute.4' Others have suggested thatthe Chapman test is the measure for the federalharmless error statute.42 Some have concluded thatthe Chapman rule is a new constitutional incursioninto matters properly the sole concern of thestates. 43 Still others have suggested that a later case,In re Winship, requires application of the Chapmanrule for all error-constitutional or not-in a crim-inal trial.

The Chapman test for harmless constitutional er-ror, regardless of its foundation orjustification, wasthe same test that Chief Justice Warren had "sug-gested" to the Connecticut court in Fahy: "Beforea federal constitutional error can be held harmless,the court must be able to declare a belief that itwas harmless beyond a reasonable doubt. 45 In

that they may, consistent with the Federal Consti-tution, be deemed harmless, not requiring the au-tomatic reversal of the conviction.

Id. at 22.9 "All 50 States have harmless-error statutes or rules,

and the United States long ago through its Congressestablished for its courts the rule that judgments shall notbe reversed for 'errors or defects which do not affect thesubstantial rights of the parties."' Id.

40Id.41 Much of Justice Brennan's dissent in Harrington v.

California, 395 U.S. 250 (1969), takes this approachalthough it does not suggest that the California rule is atissue. Field, supra note 4, contains an extensive discussionof the "overwhelming evidence" measure and its appli-cability to current harmless error cases.

42 R. TRAYNR, supra note 4, at 37-42.4 Chapman v. California, 386 U.S. at 45-46 (Harlan,

J., dissenting).44 397 U.S. 358 (1970). See Saltzburg, supra note 4, at

991. Professor Saltzburg's argument that any method of"overwhelming evidence" examination requires the ap-pellate court to use a reasonable doubt test in order tocomply with due process is persuasive. Id. at 1009-20.This position takes no account of the "class" of the error,but focuses upon the appellate fact review function anddemands that the appellate jury be bound by the sametest as the petit jury. He might have carried his pointfurther. See text accompanying notes 64-109 infra for adiscussion of the appropriateness of any appellate fact-finding in a criminal case.4" 386 U.S. at 24.

adopting the Fahy test explicitly, Justice Blackobserved that "the beneficiary of a constitutionalerror" carried the burden of demonstrating thatthe error "did not contribute to the verdict., 46

Save for the reference to the harmless errorstatutes which grew out of the law reform move-ment of the early twentieth century, Justice Blackpresented no authority for the proposition that aconstitutional violation might be harmless. Hisbrief discussion of the statutes presents no clues asto what it is about their enactment or their foun-dation which would theoretically support a harm-less constitutional error rule.47

Ironically, Justice Stewart, who dissented fromthe Court's interference with Connecticut's appli-cation of its own harmless error statute to theunconstitutional search in Fahy, and dissented fromthe Court's determination that the California com-ment rule "cost" Griffin anything he had not pre-viously spent,48 argued that the Griffin error, beingconstitutional, required automatic reversal ofChapman's conviction. Justice Stewart's Chapmanconcurrence emphasized that, despite ample op-portunity, the Court had never paused to inquireif a constitutional violation was harmful, reversingin most instances in which such a violation wasfound. He was not, however, willing to draw theharmlessness line between constitutional error andall other error. Noting that Griffin error did notpresent the appropriate vehicle for breaking with"settled precedent, 49 Justice Stewart indicated awillingness to place exclusionary rule error in the"all other" category. °

46 id.47Justice Black sat on both the Bruno and Kotteakos

Courts and was as familiar as anyone with the historyand purpose of the harmless error statutes which devel-oped out of the law reform movement. He undoubtedlyknew that the reason the statutes made no distinctionbetween constitutional error and common error was thatnobody proposing or passing the statutes ever dreamedthat anybody would ever suggest that the "minutiae"with which they were dealing would include a constitu-tional guarantee. One can only speculate that JusticeBlack felt the need of some historical support andclutched at the first arguably relevant straw.

48 Stewart's Griffin dissent was essentially a harmless-ness approach in the first instance. He could not see howtelling a jury something they already knew-that thedefendant had not testified-could matter in the result.He was, therefore, unwilling to agree that the exercise ofthe fifth amendment right cost anything. 380 U.S. at620-21 (Stewart, J., dissenting).49 386 U.S. at 45.s°Justice Stewart says that "constitutional rights are

not fungible goods" as a preface to his position that whileautomatic reversal has been the rule, the right case might

1980]

STEVEN H. GOLDBERG

Just.as Justice Stewart seemed to be willing toinclude certain constitutional error on the harmlesserror side, Justice Black seemed to concede thatcertain constitutional error requires automatic re-versal. Without embracing the precedents, Blackobserved that "prior cases have indicated that thereare some constitutional rights so basic to a fair trialthat their infraction can never be treated as harm-less error."

5'

Though Chapman may have left more questions

than it answered, it established the possibility of a

harmless constitutional error, wrapping the doc-

trine in the cloak of that harmless error doctrinewhich, in Justice Frankfurter's words, dealt withthe "minutiae of procedure."

'5 2

I. THE SNEAK THIEF

Chapman received little notice when it was de-cided because many commentators agreed pri-vately with the public assessment that "the possi-

bility that a particular error will be found 'harmless

beyond a reasonable doubt' seems slight." 53Only

Justice Harlan expected Chapman to have any im-

pact on judicial review of convictions, and his

expectation was that the federal courts would be-

come increasingly involved in what he consideredto be "state-judicial domains."' Both predictions

present a place for harmless error. His only example, andobvious choice for the place, is the exclusionary rule.While he speaks in terms of a constitutional right, it isclear that he does so by force of precedent rather thanpersonal conviction. His footnote to the subject leads oneto believe that he is more interested in returning theexclusionary rule to the realm of common error than heis in creating a harmless constitutional error rule. Id. at44 n.2.

s' 386 U.S. at 23. Justice Black cites three examples,id. at 23 n.8: Gideon v. Wainwright, 372 U.S. 335 (1963)(right to counsel); Payne v. Arkansas, 356 U.S. 560 (1958)(coerced confession), and Tumey v. Ohio, 273 U.S. 510(1927) (impartial judge). Justice Stewart cites additionalinstances in which automatic reversal is indicated. JusticeBlack gives no reason for those he selects, does notnecessarily endorse the result, and provides no clue as towhat characteristic unifies the cases past his "so basic toa fair trial" language. Given his total incorporation ap-proach to due process, see Adamson v. California, 332U.S. 46 (1947) (Black, J., dissenting), it is unlikely thathe means to suggest that there is a constitutional differ-ence among rights found within the first eight amend-ments. His citation, like Justice Stewart's, may be theresult of frustration with the process of trying to set acode of procedure from the Bill of Rights, and striking areasonable balance between clear rules and flexible ap-plication.

52 Bruno v. United States, 308 U.S. 287, 294 (1939)."3 The Supreme Court, 1967 Term, 81 HARV. L. REv. 205

(1967).54 386 U.S. at 57 (Harlan, J., dissenting).

were wrong. The "slight" possibility became acommon occurrence in both the state and federalcourts, 5s and the federal courts did not use Chapman

as a tool for fixing their own notion of "harm"upon state courts with contrary views. There is

some irony in the fact that, on the 6ontrary, Chap-

man became the magic formula by which the state

courts found harmless constitutional error in recordnumbers-findings to which the federal courts gave

substantial deference.56

Although Chapman remains the most commonly

cited authority for the proposition that any partic-

ular constitutional error is harmless, it may not beaccurate to ascribe the significant evils of the doc-

trine to the Chapman opinion. Although Chapman

introduced the harmless constitutional error, thedoctrine as presented possessed only the potential,

not the doctrinal structure, for constitutional and

institutional mischief. Traynor thought the Chap-

man test to be so stringent as to be tantamount to

an automatic reversal rule.57 Had the Supreme

Court interpreted and the other appellate courtsapplied Chapman to allow a finding of harmlessnessonly after an appellate court had examined theerror alone and found beyond a reasonable doubt

that it could not have contributed to the verdict,

the predictions of inconsequence might have

proven correct. Cases decided after Chapman re-laxed the rigor of the test and applied it to circum-

stances which could not have been contemplated,

and indeed, would have been disavowed by the

Chapman majority.Any new doctrine or exception to an established

doctrine must be expected to grow past the param-

eters set in the decision which created it. Theharmless constitutional error doctrine has a char-

' The data from Harmful Use, supra note 3, at 544-48,on the incidence of harmless error indicate the federalcourts perhaps were as anxious as the state courts to usea mechanism which would allow them to find a consti-tutional error and still affirm the conviction.

' Because a finding that an error is harmless is afactual determination, there is little realistic opportunityfor review by the federal courts of a state appellate courtdecision. An interesting exception is the kind of casewhere the state appellate court finds an error harmlessand the federal court need not second guess the stateappellate court's reading of the statute in order to findthat there was harm in the error. Allison v. Gray, 603F.2d 633 (7th Cir. 1979), is an example of such a case.The Wisconsin Supreme Court found that the failure toallow a defendant to present alibi testimony was error,but harmless. Allison v. State, 62 Wis. 2d 14, 214 N.W.2d437, cert. denied, 419 U.S. 1071 (1974). The Seventh Circuitreversed, noting that the Wisconsin court could not pos-sibly judge the weight or effect of alibi evidence that wasnot in the record.

17 R. TRAYNOR, supra note 4, at 43.

[Vol. 71

HARMLESS ERROR." CONSTITUTIONAL SNEAK THIEF

acteristic unique among doctrines with constitu-tional impact that makes such growth immediate,indiscriminate, and unpredictable. It has no sub-stantive doctrinal base. As a result, the doctrine hasbecome a major thief of constitutional rights with-out any particular notice or analysis. Unlike thesubstantive constitutional decisions such as Mapps5

Miranda,5 9 and Wade,60 or even the procedural con-stitutional decisions such as Linkletter v. Walker,61

harmless constitutional error carries with it no legalissue of primary concern to an appellant. Appel-lants are unable to launch a consistent and effectiveattack against the doctrine because no single ap-pellant has any doctrinal stake in the issue. Eachappellant hopes only to avoid the factual trap ofharmless constitutional error. He has little oppor-tunity to challenge on the factual issue, let aloneon the propriety of the doctrine.62 Consequently,the courts failed to provide analysis in the devel-opment of harmless constitutional error. That fail-ure has created an appellate procedural doctrinewhich has caused "mischief'6 beyond anyone'sexpectations. The doctrine has created appellatefactfinding which denies the constitutionally guar-anteed right to trial by jury. In addition, thedoctrine erodes constitutional principles at alllevels of the criminal justice system from prosecu-tion to Supreme Court review without ever afford-ing an opportunity for a hearing on the merits ofthe principle eroded.

III. "... A FAMILIAR STANDARD"

Justice Black's observation that the reasonabledoubt standard was a familiar one to appellate

5 Mapp v. Ohio, 367 U.S. 643 (1961) (searches).69 Miranda v. Arizona, 384 U.S. 436 (1966) (confes-

sions).6 United States v. Wade, 388 U.S. 218 (1967) (lineuP 381 U.S. 618 (1965) (retroactivity).

62 The nature of the harmless error determination issuch that an appellant cannot argue its inapplicabilitywithout an implicit concession that the inquiry is justifiedand without taking valuable time and energy away frompersuading the court that there is error in the firstinstance. Further, few appeals proceed with the fullrecord available to the appellate court or sufficient timefor the appellant to adequately parse the record to estab-lish the harmlessness of an error that the appellate courthas not yet found.

6Justice Black anticipated that harmless error rulescould create "mischievous results when, for example,highly important and persuasive evidence, or argument,though legally forbidden, finds its way into a trial inwhich the question of guilt or innocence is a close one."386 U.S. at 22. He did not anticipate the mischief whichthe doctrine might bring to the appellate process or toconstitutional guarantees.64 "While appellate courts do not ordinarily have the

courts identifies precisely the problem which aroseafter the Supreme Court decided Harrington v. Cal-ifornia.65 Appellate courts never operate as fact-finders for matters which must be determined be-yond a reasonable doubt. They rarely operate asfact-finders at all. 66 Most importantly, until theCourt approved the "overwhelming evidence" ap-proach to harmless constitutional error, no courthad the power to enter a guilty verdict on its ownjudgment when the defendant properly exercisedhis right to a trial by jury.67

Because Chapman seemed to reject the Californiacourt's "overwhelming evidence" approach toharmless error, and to approve a test that requiredthe court to focus instead on whether the errorcontributed to the verdict, Glenn Harrington'smurder conviction would likely not stand. Har-rington, a caucasian, and three blacks were jointlytried and convicted of murder. Each of Harring-ton's alleged accomplices confessed. Their confes-sions were admitted at the trial in violation of therule of Bruton v. United States, s prohibiting theintroduction, against a codefendant, of another'sconfession when that other individual did not takethe stand and thereby submit to confrontation byhis codefendant. It is difficult to imagine how anycourt, focusing only upon the illegally admittedconfessions, could determine beyond a reasonabledoubt that the confessions of the accomplices didnot contribute to Harrington's conviction-partic-ularly when those confessions described the fourthparticipant as "the white guy."9 Harrington made

original task of applying such a test, it is a familiarstandard to all courts .... "Id. at 24.

6 395 U.S. 250 (1969).66 Rule 52 of the Federal Rules of Civil Procedure is

exemplary of the reticence of appellate courts to review,let alone make initial determinations of fact. "Findingsof fact shall not be set aside unless clearly erroneous, anddue regard shall be given to the opportunity of the trialcourt to judge the credibility of the witnesses." FED. R.Civ. P. 52(a).

67 The appellate court's action in finding an errorharmless by assessing the weight of the remainder of theevidence is not, in a technical sense, the direction of averdict nor the granting of a judgment n.o.v. to thegovernment. However, it is a determination of guilt uponfacts which have never been considered by a jury andwhich, given the appellate court's finding of guilt, neverwill be.

68 391 U.S. 123 (1968).69 None of the defendants mentioned Harrington by

name. The Court, however, accepted Harrington's argu-ment that, when the defendants said "the white guy,"there was not much doubt who the speaker had in mindas Harrington sat in the dock with three blacks. 395 U.S.at 253.

STEVEN H. GOLDBERG

statements which, according to the Court, "fellshort of a confession but which placed him at thescene of the crime." 70 According to Justice Douglas,writing for a five-member majority, those state-ments and other evidence were "so overwhelming"that the admission of the co-defendant'sconfession was harmless beyond a reasonabledoubt.

7 1

The Harrington majority, which included JusticeBlack, the author of Chapman, stated emphaticallythat they did not "depart from Chapman; nor ...dilute it by inference., 72 Justice Brennan, joinedby ChiefJustice Warren and Justice Marshall, tookexception, contending that Chapman had been over-ruled: "The Court today by shifting the inquiryfrom whether the constitutional error contributedto the conviction to whether the untainted evidenceprovided 'overwhelming' support for the convic-tion puts aside the firm resolve of Chapman .... 73

Harrington was followed by the remarkably simi-lar case of Schneble v. Florida.74 Although the dissen-ters, including Justice Douglas, the author of Har-rington, complained bitterly that Schneble's relianceupon Harrington was misplaced, the majority usedthe "overwhelming evidence" approach that Jus-tice Douglas had adopted for the Harrington facts. 75

In addition, Justice Rehnquist's pointed use of"probability" 76 language from the Harrington deci-sion, plus his own peculiar phrasing of the Chapmanholding,77 left the burden of proof unclear, both as

to what it was and who was to carry it. AfterSchneble, one may argue reasonably that an appel-late court faced with an error of constitutional

70Id. at 252.

71 Id. at 254.72 id.

73 Id. at 255 (Brennan, J., dissenting).74 405 U.S. 427 (1972).75 "Not only is the independent evidence of guilt here

overwhelming, as in Harrington.... " Id. at 431.76 Although it is unlikely that Justice Douglas intended

to change the test from beyond a reasonable doubt to oneof probability, he used language in Harrington whichJustice Rehnquist was able to use to support such achange: "Our own reading of the record and on whatseems to us to have been the probable impact ... on theminds of an averagejury." Id. at 432 (quoting Harringtonv. California, 395 U.S. at 254).77 Justice Black put the burden of demonstration uponthe prosecution, "requiring the beneficiary of a constitu-tional error to prove beyond a reasonable doubt that theerror complained of did not contribute to the verdictobtained." 386 U.S. at 24. Justice Rehnquist implied thatthe demonstration must be made by the defendant:"Thus, unless there is a reasonable possibility that theimproperly admitted evidence contributed to the convic-tion, reversal is not required." 405 U.S. at 432.

dimension should make its own independent eval-uation of the evidence, and reverse the convictiononly if it was persuaded that the average jurywould have changed its verdict had the illegalevidence been excluded. 78 By a subtle rearrange-ment of words and ideas, Justice Rehnquist con-verted a test which forced the prosecution to showbeyond a reasonable doubt that the error did notcontribute to the verdict, into a test which forcedthe defendant to show that the error was of suchsignificance that without it the defendant wouldbe entitled to a directed verdict of acquittal.7 9

Most of the commentary concerned with harm-less constitutional error has focused upon subclassesof constitutional rights which, in the minds of thecommentators, ought to be exempt from harmlesserror analysis.80 These commentators focused onthe nature of the constitutional right violatedrather than the nature of the process by which theharmlessness is determined.8 For the most partthey follow Justice Black's position and have con-centrated on constitutional error which affects the"process" of the trial, such as failure of counsel ordiscrimination in the selection of a jury panel,rather than on those errors which affect the amountof evidence presented. They have argued that whena violation of the right is either harmless by defi-nition-jury accessS2-or beyond precise determi-nation-effective assistance of counsel84-harmless

78 ChiefJustice Burger's opinion in Holloway v. Arkansas,

435 U.S. 475 (1978), rejects the contention that harmlesserror analysis is appropriate when a defendant raises afailure of effective assistance of counsel due to a conflictof interest. His rejection is based upon the propositionthat "a rule requiring the defendant to show that aconflict of interests ... prejudiced him ... " would notbe susceptible to evenhanded application as is requiredin harmless error analysis. There can be little doubt fromhis explanation that he considers the defendant to havethe burden of demonstration with respect to harmlesserror.79justice Rehnquist's statement that "[i]n this case,we conclude that the 'minds of an average jury' wouldnot have found the State's ease significantly less persua-sive had the testimony as to Snell's admission beenexcluded," can only be read in the context of his opinionto mean that the jury verdict would not have beendifferent. 405 U.S. at 432. The effect is to make theharmless constitutional error test the equivalent of aguilty or not guilty determination, allowing only thosewho can demonstrate to the appellate court that they arenot guilty as a matter of law-that is, can demonstrate areasonable doubt as to the entire case-to gain a newtrial when constitutional rights are abridged.

See, e.g., Mause, supra note 4.o See, e.g., Note, Harmless Constitutional Error, 20 STAN.

L. REV. 83 (1967).82 Field, supra note 4, at 19-20.8 Mause, supra note 4, at 541.

[Vol. 71

HARMLESS ERROR" CONSTITUTIONAL SNEAK THIEF

constitutional error analysis is inappropriate. Nonehas suggested that the harmless constitutional errorprocess itself violates the defendant's right to a trialbyjury.'

When an appellate court tests for harmlessnessby reviewing the record to determine whether theremainder of the evidence is so overwhelming thatthe error did not contribute to the verdict, it sits asan appellate jury.85 The test assumes that the errormade a difference in the amount of evidence pre-sented to the jury.86 When the appellate court

8' Professor Saltzburg has suggested that the Chapmanrule is required for all error in a criminal case-be itconstitutional or common-by virtue of the due processclause and the right to trial by jury. Saltzburg, supra note4, at 991-93. Professor Mause briefly raised the possibilitythat the jury trial guarantee made any harmless consti-tutional error rule unconstitutional, but rejected theproposition because nothing requires a state to provideappellate review of any kind in a criminal case. Mause,supra note 4, at 531-32. Professor Saltzburg raised thesame problem for his proposition that the jury trial rightprecluded any test but Chapman's reasonable doubt testfor all error in criminal trials. Saltzburg, supra note 4, at1028. Saltzburg was probably correct with respect tocommon error. Mause was undoubtedly wrong with re-spect to constitutional error. States cannot avoid federalreview, upon habeas corpus, for error of a constitutionalnature. It follows that any jury trial problem resultingfrom application of a harmless error rule cannot be setaside for constitutional error, although it may be forcommon error.

85 Professor Field, supra note 4, at 39, suggests that theHarrington test is not really "overwhelming evidence" but"cumulative evidence." Whatever merit there may be tothe distinction, the appellate court must, in the firstinstance, make ajudgment about the quantity or qualityof the remainder of the evidence. Professor Saltzburg,supra note 4 at 1014, n.89, argues that there is no practicaldifference between the effect of the evidence test and theoverwhelming evidence test so long as the appellate courtuses a "virtually certain" or beyond a reasonable doubttest on the remainder of the evidence. He maintains thatin any instance where the overwhelming evidence testleaves the Court with the belief that the defendant isguilty beyond a reasonable doubt it would also find thatthe error had made no contribution to the verdict. Whilethat may often be the case, it is not necessarily so. Forexample, in Allison v. Gray, 603 F.2d 633 (7th Cir. 1979),the evidence would leave the average skeptic convincedof the defendant's guilt. However, the error was failureto allow the defendant to present alibi evidence. Underthe effect of the error test, the exclusion of the defendant'sevidence could never be harmless. Regardless of whetherit would often lead to the same result, the appellateactivity in the effect of the evidence test is a traditionalappellate exercise of determining a factual "could" byassuming all facts in a light most favorable to the partywithout the appellate burden, while in the overwhelmingevidence test the court is reaching an original fact judg-ment.

8o Professor Field suggests that the overwhelming evi-

decides that the evidence presented to the juryminus (or plus) the evidence produced (or ex-cluded) by the error, it is the only factfinder whichhas ever made a guilty/not guilty judgment aboutthe new amount of evidence.

An appellate court's determination that a set offacts supports a guilty verdict, in the first instance,is unprecedented. The failure of precedent reflectsa consistent and long tem judgment about thewisdom and constitutionality of such an appellatedetermination. For reasons so familiar that JusticeBrennan did not think repetition was necessary inHarrington, appellate courts restrict their factfindingactivity to the review of what a previous factfinderfound. 7 The review is limited to determinationsthat the factfinder's conclusion cannot be justifiedeven when all of the evidence is viewed in a lightmost favorable to the conclusion.88

An appellate court defies common sense when itsteps out of its traditional role as a reviewing courtand attempts to operate as a primary factfinder.Appellate review of an entire trial transcript is anincredibly inefficient use of appellate court time.To pursue such a course in order to determinewhether error is harmless, so that judicial economymight be served is not only ironic, it is nonsensi-cal.89 In addition, unless the litigants are placed on

dence test may be applicable to constitutional errors attrial which do not affect the amount of evidence pre-sented to the jury. Field, supra note 4, at 16. None of theSupreme Court decisions finding harmless constitutionalerror has involved an error which did not affect theamount of evidence. Those cases to reject harmless con-stitutional error treatment for particular error have in-volved errors which did not quantifiably affect theamount of evidence. While the necessary speculationinvolved with such analysis has never been assigned as areason for excluding all such error from harmlessnessanalysis, it seems to be the rationale from case to case.See, e.g., Holloway v. Arkansas, 435 U.S. 475 (1978).

87 395 U.S. at 257 (Brennan, J., dissenting). See discus-sion at text accompanying notes 91-98 infra, for the oft-cited factors militating against appellate factfinding.

88 Justice Rehnquist misused the familiar sufficiency ofthe evidence test in Schneble v. Florida, 405 U.S. 427.He relied upon Rogers v. Missouri Pacific Ry. Co., 352U.S. 500 (1957),'to support a factfinding presumption infavor of the government on an issue for which it had theburden of proof. 405 U.S. at 432. Justice Marshall, indissent, wasted no time in pointing out that the appellatefunction in determining the sufficiency of the evidence ina civil case was exactly the opposite of the function setout by the harmless error test, and employed a presump-tion exactly opposite of that underlying the reasonabledoubt burden. 405 U.S. 433-34.

89Justice Stewart, concurring in Chapman, alludes tothe difficulty of an appellate court reviewing each tran-script from case to case in order to determine harmless-

1980]

STEVEN H. GOLDBERG

notice that the error is conceded and harmlessnessis the only issue, the appellate court must reviewthe entire transcript and make a determinationwithout the benefit of the litigants' view of thefacts-a substantial departure from the normalmethod of appellate review. To the extent thatappellate time is saved by the litigants' shapingand narrowing the issues, the inability of the liti-gants to contribute to the-appellate decisionmakingprocess harms the litigants and wastes court time.Unless the courts adopt a policy of total fact reviewfirst and legal issue second,9° the appellate courtcannot begin the time-consuming task of fact re-view until it has taken the time to perform itstraditional function of reviewing for and, in this

case, finding error. From the perspective of efficientappellate practice only-judicial economy of akind-the harmless constitutional error doctrineinvolves a tremendous increase in appellate courttime and a diminution of the traditional assistancethat counsel provides.

From an institutional perspective, the time costis not as substantial as the cost injustice. Appellate

courts are, by their position as dispassionate andremoved arbiters of the law, extremely poor finders

of fact. Appellate courts' deference to trial courtfactfinding is not a matter of accident. 9 A cold

record, assuming that it is accurate, cannot substi-tute for a trial. Every trial lawyer knows that the"facts" of demeanor are at least as important as

the "facts" of testimony. An appellate court read-ing a record in its entirety92 knows nothing of the

unreasonable pause, the inappropriate smile, thesarcasm that changes a "sure" which means "yes"

ness, and then candidly admits that "[t]his burdensomeobligation is one that we here are hardly qualified todischarge." 386 U.S. at 45.

90 In Milton v. Wainwright, 407 U.S. 371 (1972), overthe vigorous dissent of four justices, Chief Justice Burgerstood appellate procedure on its head. He assumed theerror-the real question in the case-and proceeded, sohe said, to review the entire record to make a harmlessnessdetermination.

9i "Such deference is particularly apposite because thetrial judge and jury are closest to the trial scene and thusare afforded the best opportunity to evaluate contradic-tory testimony." Mincey v. Arizona, 437 U.S. 385, 408(1979) (Rehnquist, J., dissenting).

9 There is no way to insure that judges read the entiretranscript. Given the volume in most appellate courts, itstretches the imagination to believe that even one of thejudges would read the entire record. In fact, most matterscome to appellate courts with abbreviated records col-lected in an appendix by litigants to whom harmlessnesshas not occurred. The appendix and record are chosenfor relevance to the legal issues raised. Good practice andappellate rules encourage the abbreviated record.

to a "sure" which means "I don't believe that" or"I don't agree." Appropriately, every trial courtinstructs the jury that it is the sole judge of witnesscredibility. Rule 52's admonition that appellatecourts in civil cases shall give "due regard ... tothe opportunity of the trial court to judge thecredibility of the witnesses"93 is no accident. Oneof the problems with appellate factfinding is thatthe appellate court is likely to be wrong.

The greatest cost of the harmless constitutionalerror9 rule is its usurpation of the jury function.Aside from the likelihood of erroneous factfindingbecause a cold record is a poor communicator, anappellate court is far removed from being a jury,and jury trials comprise the heart of our criminaljustice system. We are probably better off withjuries making "wrong" decisions than with judgesmaking "right" ones.95 The decisionmaking gulfbetween three appellate judges reading a transcriptat their leisure in chambers and twelve citizenslocked in ajury room is not bridged by an appellatejudge's intellectual understanding of the reasona-ble doubt standard. In comparing the appellatejury to the petit jury some questions should beasked. Which defendant, let alone the defendant'slawyer, would demand ajury trial and then allowan appellate judge to sit on the jury? Even withthe limited voir dire available in most federalcourts, if a lawyer were foolish enough to seat anappellate judge on a jury, would the lawyer do so

93 FED. R. Civ. P. 52(a).9 Justice Harlan, in his Fahy dissent, points out that a

nonconstitutional error can be as prejudicial to a defend-ant as a constitutional error, and in some instances, moreso. 375 U.S. at 94. Professor Saltzburg, supra note 4, at989-90, makes the same point in arguing for the sameharmless error rule in criminal cases involving constitu-tional and common error. While the automatic reversalrule for harmless constitutional error does not suffer fromthe same vulnerability to the "appellate review not re-quired" argument as does harmless common error, thepotential for denial of a jury trial exists as strongly forcommon error as it does for constitutional error when thecharacter of jury justice (as opposed to appellate justice)is at issue.

9 Justice Harlan, in a rare example of poor judgment,once suggested that "untrained jurors are presumablyless adept at reaching accurate conclusion of fact thanjudges .... " Duncan v. Louisiana, 391 U.S. 145, 188-89(1968) (Harlan, J., dissenting). Even if true, the value incitizen participation may outweigh the value of a deci-sionmaking system which makes more correct decisions.In the law generally, and in criminal law particularly,the societal acceptability of the decision may be moreimportant than its correctness. Juries represent an insti-tutional insurance policy for the continued acceptabilityof the decisionmaking system.

[Vol. 71

HARMLESS ERROR: CONSTITUTIONAL SNEAK THIEF

without some voir dire? Would the lawyer at leastfind out who thejudge was before failing to exercisea peremptory?96 Assuming a lawyer who would trya factual defense to a jury of appellate judges,would he agree to waive summation?9 7 Would thatlawyer allow the appellate jury to know that aprevious jury found the defendant guilty? Wouldthe lawyer agree that group decisionmaking meansnothing and that the jury members could go backto their offices and.read through notes at their ownpace until they reach a decision? Would the lawyeragree that jury nullification was of no particularvalue to the defendant?9s Any lawyer who mightagree to all of the above would not have muchquarrel with a three-person jury finding the de-fendant guilty, two-to-one.

Duncan v. Louisiana,9 decided a year after Chap-man, held that the right to a trial by jury was oneof those fundamental rights which not even theexistence of state boundaries could destroy. Pre-sumably it would be one of those rights "so basicto a fair trial"'" that even Justice Black wouldconsider harmless error analysis to be inappro-priate. In deciding that a trial judge's decision thatGary Duncan was guilty of simple battery beyonda reasonable doubt could not be squared with thesixth amendment right to ajury trial, Justice Whitereviewed the history of American juries. He as-signed "oppression by the Government" and "pro-tect[ion] ... against judges" as two of the majorreasons for the insistence upon the right to jury

9A danger exists that when harmlessness is at issuethe appellate lawyer who decides to argue harm is makinghis pitch to the jury's "boss," not the jury. If longtranscripts are to be read, clerks are likely to do thereading, not judges.97 Because harmlessness is rarely at issue in briefs or at

argument, the defense lawyer never gets to argue thefacts and the law to the appellate jury. Even if theopportunity is there, the judges have not yet read therecord, and the lawyer is arguing from factual thin air.Additionally, in criminal cases much of the argument forthe defendant is related to the burden of proof. There isno opportunity to make that argument, and there is somereason to believe that appellate judges are in more needof the refresher than some juries. See, e.g., Milton v.Wainwright, 407 U.S. 371 (1972).

98Jury nullification is not a particularly commonevent, and may not often be affected by an evidenceerror. However, there may be circumstances in which ajury failed to exercise its power to nullify the law becausethe error admitted evidence that dissuaded it from nul-lification, or excluded evidence which, if heard, wouldhave persuaded the jury to nullify.99 391 U.S. 145 (1968).10o Justice Black's description of those errors which the

Court had found to require automatic reversal. 386 U.S.at 23.

trial in criminal cases.1 'x

After Duncan, the Court spent some time decidingexactly how many people were needed for a jury,what constituted an impartial jury,'0 2 and howmany of those impartial jurors had to agree beforea criminal verdict was proper. While there wassubstantial disagreement, no Justice on the Courtever suggested that three is enough to make ajury,that less than six votes is enough to convict, or thatappellate judges are diverse enough to constitute acommunity cross section. Although Williams v. Flor-ida'0 3 held that the jury guaranteed by Duncan neednot be made up of twelve persons, and Apodaca v.Oregon10 4 allowed that a criminal jury need not beunanimous, Burch v. Louisiana'0 5 made it clear thatthe sixth amendment required a jury as small assix persons to be unanimous before it could returna guilty verdict.' °6 Furthermore, Taylor v. Louis-iana'0 7 and Duren v. Missouri' established that thedefendant's sixth amendment right to an impartialjury meant that the jury must be drawn from afair cross section of the community. An appellatecourt of three, deciding guilt by a vote of two-to-one is hardly a jury.

The effect of the decisions in Harrington andSchneble, and their approval of the overwhelmingevidence approach to harmless constitutional erroris to deny the defendant/appellant a right to atrial by jury. The appellate court, by definition,sits as a jury and makes a guilt determinationbased upon an amount of evidence upon which nojury has passed. While double jeopardy in theclassic sense is not involved, a certain discomfiturelies in an appellate court's determination that alower appellate court erred in its judgment that anerror was not harmless.'0o Lastly, there is, at aminimum, some embarrassment in a decision

'0' 391 U.S. at 155-56.'02 The sixth amendment guarantees, among other

things: " ... trial, by an impartial jury." U.S. CoNrsr.amend. VI.

103 399 U.S. 78 (1970) (approving a jury of six).104 406 U.S. 404 (1972) (approving a guilty verdict by

ten-to-two jury votes).05 441 U.S. 130 (1979).106 Ballew v. Georgia, 435 U.S. 223 (1978), somehow

discerned a constitutional difference between six and five,allowing a jury of six, but not of five.

'07 419 U.S 522 (1975).108 439 U.S. 357 (1979).'09 The dissent in Milton v. Wainwright, 407 U.S. at

378, pointed out that no less than four other courts hadconsidered the matter and none had found the error tobe harmless. If each used the "overwhelming" evidencetest, only the Supreme Court-five-to-four-believedthat the untainted evidence proved guilt beyond a rea-sonable doubt.

1980]

STEVEN H. GOLDBERG

which says that an error was harmless beyond areasonable doubt: five-to-four.

IV. SECOND STORY JoB ON THE CONSTITUTION

The harmless constitutional error doctrine worksonly a petty theft on individual defendants' rightsin specific cases but its consistent application exertsa more profound effect upon society. The harmlessconstitutional error rule, regardless of the test, mil-itates against basic freedoms and controls upongovernmental institutions that operate against in-dividuals. At all levels of the criminal justice sys-tem, the harmless constitutional error rule dilutesthe impact of most constitutional criminal proce-dure decisions of the last quarter century whichpreserve individual rights against contrary claimsof necessity by the government. The rule allows anad hoc, after the fact, factual judgment by anappellate court which makes it difficult for anindividual to attack it and diminishes the level ofprotection provided by specific constitutional pro-visions without affording any opportunity to arguethe issue's merits. If, as some claim, many of theWarren Court decisions of the 1960s were ill-ad-vised, if the society has, indeed, been damaged bythe change in the relationship between the individ-ual and the state as incarcerator, that is a matterto be addressed on the merits, not through theprocedural backdoor of harmless constitutional er-ror.

Harmless constitutional error is particularly in-appropriate because of the special relationship be-tween the Constitution and the courts. All of theassumptions necessary to support any harmlesserror doctrine are inapplicable when the error atstake abrogates a constitutional right. Any doctrinewhich allows an appellate court to identify an errorand then find it "harmless" assumes that the ap-pellate function is limited to the correction of aninappropriate trial court result. Further, it assumesthe decision will affect the litigants alone."0 Theharmless error doctrine assumes that any otherinterest, other than the correction of an inappro-priate trial result, is sufficiently insignificant tobalance unfavorably against the nonconstitutionalvalue ofjudicial economy."'

1o While the result might affect future and unknownlitigants, it is the litigants in the case or those that maysomeday be similarly placed that are affected, to theexclusion of the public in general.

" A public interest exists for every law, including theMissouri constitutional requirement which caused thereversal of Campbell's case for want of a "the" in theindictment. State v. Campbell, 210 Mo. 202, 109 S.W.

At least since Marbury v. Madison, 2 the SupremeCourt obligation of judicial review has been thelifeblood of the living Constitution. Justice Har-lan's Chapman dissent decries what he sees as theCourt's "assumption of what amounts to a generalsupervisory power over the trial of federal consti-tutional issues in state courts," 113 but he is a lonevoice. Justice Black left little doubt that federalismconcerns are insufficient to justify state interferencewith rights emanating from the federal Constitu-tion.1

4

The special relationship between the Constitu-tion and the Court, in which each gains its strengthfrom the other, is unique when the rights in ques-tion are those enumerated in the first eight amend-ments. One does not have to accept Justice Black'stheory of total incorporation 115 to recognize thatthe first eight amendments are antimajoritarianand antigovernment. In the American system, thecourts historically have been the only institutionsufficiently separated from the political system toact as the preserver of those antimajoritarian andantigovernmental rights. The cost of that preser-vation has never been small: "The criminal is togo free because the constable has blundered."

116

If certain errors are to be susceptible to harmlesserror treatment and others are not," 7 the Consti-

706 (1908). The harmless error rule assumes that what-ever that interest may be it is of less weight than theinterest in judicial economy.

112 5 U.S. (1 Cranch) 137 (1803).113 386 U.S. at 46-47 (Harlan, J., dissenting).114 See note 37 supra.lnjustice Black argued that the first eight amend-

ments were totally incorporated into the due processclause of the fourteenth amendment and were therebyapplicable to the states. His position, most fully set out inhis dissent in Adamson v. California, 332 U.S. 46, 68(1947) (Black, J., dissenting), never held a majority of theCourt. Ironically, his position was set out in oppositionto the due process approach of Cardozo, Frankfurter,Harlan and others, which inquired as to whether thematter at issue was fundamental to the American conceptof criminal justice. His language in Chapman, "so basic toa fair trial," comes perilously close to the Cardozo-Frank-furter-Harlan concept of distinguishing between rightsbased upon "fundamentalism."

"'People v. DeFore, 242 N.Y. 13, 21, 150 N.E. 585,587 (1926) (opinion of Cardozo, J.).

"'Justice Black, by his Chapman acknowledgment ofthe Court's prior decisions, Justice Stewart, by his desireto exclude the fourth amendment, and the many com-mentators by their various selections have all agreed thatthere are some constitutional errors which ought to besubject to the harmless error approach and others whichshould not. None has presented a particularly persuasivereason for discrimination between constitutional errors.

[Vol. 71

HARMLESS ERROR: CONSTITUTIONAL SNEAK THIEF

tution is the only reasonable place to draw theline. 8 The efficacy of that line is based upon theassumptions of both the harmless error doctrineand the system of constitutional jurisprudence.There are three major differences between noncon-stitutional and constitutional rights which demanddistinction between them when the appropriate-ness of harmless error analysis is at issue.

First, nonconstitutional rights, the abrogation ofwhich create trial error, are transitory and political.They are legislatively created, maintained, 'andchanged. Their preservation is ensured by the ul-timate safeguard of majority approval, or at leastpolitical process approval. Constitutional rights, onthe other hand, are immune to the political process,at least as to preservation." 9 Legislators and elec-torates have no function in the preservation ofconstitutional antimajoritarian and antigovern-mental individual rights. The courts, rightly orwrongly, are the sole institution for maintenanceand change.

120

Second, nonconstitutional rights likely to be im-plicated in trial error are, generally, of a kind for

"8justice Harlan compellingly argues that the Chap-man decision, to the extent it conceives "of an applicationof harmless-error rules as a remedy designed to safeguardparticular constitutional rights," 386 U.S. at 50 (footnoteomitted) (emphasis added), is unsupportable. As a deviceto insulate only some constitutional rights from applica-tion of state harmless error rules, a federal harmless errorrule has little to recommend it. Neither Black nor Stewartsuggests a rationale for the application of the stricter testto some constitutional rights. Once the judgment is madethat all constitutional rights are beyond harmless consid-eration, Harlan's complaint about a failure of reason isgone. Perhaps he would disagree with a result which putall constitutional error beyond the reach of harmlesserror, however, clearly he agrees with the propositionthat the nature of the Court's relationship to the Consti-tution makes the question of the validity of a harmlesserror rule a federal constitutional question unaffected bythe nature of the specific constitutional right.

"9 Concededly, constitutional rights are subject to thepolitical process through the power of amendment, butthe political activity relates to repeal and addition ratherthan preservation. The political process possibly coulddestroy a constitutional right such as the fourth amend-ment, but the political process of constitutional amend-ment probably could not preserve the right.

120 In the fourth amendment context, a number ofcommentators have suggested that the executive mightprotect and preserve the privacy of citizens more effec-tively than the exclusionary rule. See, e.g., Kaplan, TheLimits of the Exclusionay Rule, 26 STAN. L. REv. 1027(1974). While there might be a chance, it is a little likeputting the foxes in charge of the chicken coop. It in-dulges the assumption that there is no need for theantigovernment provision in the first instance.

which society has little, if any, interest outside ofthe conduct of the trial.' 2' Constitutional rightsoften implicate substantial societal interests exclu-sive of the error in a particular trial. In those caseswhere the defendant stands as society's surrogate-search cases, for instance-society's interest in theright has nothing to do with the trial result. Thesociety's interest in the right is furthered, or not,by the result of the suppression hearing.1"'

Finally, nonconstitutional rights which are ab-rogated by trial error are usually neutral. They areas likely to behefit the government in a given trialas they are the defendant. Hearsay, for example,may benefit the government in one instance andthe defendant in another. Constitutional rights,such as the right to confrontation, which might beabrogated by a trial error, are always beneficial tothe defendant and consistently restrictive with re-spect to the government's case. Each of these threedifferences has institutional consequences whichdifferentiate common error from constitutional er-ror when the appropriateness of a harmless errordoctrine is the issue. The harmless constitutionalerror doctrine has interfered with the SupremeCourt's task of interpreting the Constitution, andmay have, in one instance, changed other interpre-tations without notice.

Harrington and Schneble were both "confronta-tion" cases and represent a line of cases whichdemonstrates two ways in which the harmless con-stitutional error doctrine interferes with the orderlydevelopment of constitutional law. Bruton v. UnitedStates," decided one year after Chapman and beforeHarrington, overruled Delli Paoli v. United States."'Delli Paoli allowed the confession of a codefendantin a joint trial to be offered despite the fact thatthe individual did not testify. The Court held aninstruction to the jury that the confession could beconsidered only against the confessing codefendantsufficiently protected the other defendant's right toconfrontation. The Bruton Court, speaking throughJustice Brennan, held that the instruction to dis-regard would not suffice in protecting the other

121 Obviously, some state rules affect conduct outsidethe scope of the trial, for example, a requirement that aconfession be signed. Most nonconstitutional trial errors,however, will involve rules designed for the conduct oftrials.

"'2 This is not intended to imply that society has nointerest in the trial result, but only that it has a distinctinterest in the fourth amendment protection. It is thatdistinct interest which is not within the assumptionsunderlying the harmless error rule.

1" 391 U.S. 123 (1968).A 352 U.S. 232 (1957).

STEVEN H. GOLDBERG

defendant's right to confrontation. In reaching thatconclusion, Brennan made an observation which,though true, was ignored the next year in Harring-ton: "We, of course, acknowledge the impossibility

of determining whether in fact the jury did or didnot ignore [the] statement inculpating petitioner indetermining petitioner's guilt. ' ' 125 Although bothDouglas' majority opinion and Brennan's dissentin Harrington concentrated on the effect of thedecision on Chapman, Harrington changed what ap-peared to be the basis for Bruton: The court cannever be sure of the jury's response to a codefen-dant's confession, and, therefore, it will take no

chances that the defendant's confrontation rightmight be abridged. While Harrington did not over-rule Bruton, it certainly did for all "good" govern-ment cases. Harrington's reliance upon "overwhelm-ing evidence" can only mean that the Bruton con-frontation right is unnecessary for the apparentlyguilty. In Harrington's situation, a statementwhich could be construed to implicate was held tofall short of a confession.

126

In Schneble, the petitioner's "confession" 127 wasmore inculpatory than Harrington's, though therewas some question about its reliability.' 28 JusticeMarshall, in dissent, attempted to show that Schne-ble's reliance upon Harrington was misplaced be-

cause Harrington involved a codefendant's confes-sion which was "merely cumulative" of the defend-ant's "confession" to having been at the scene ofthe crime. However unpersuasive that distinctionmight have been, Justice Marshall said propheti-

cally what might just as well have been said in

'25 391 U.S. at 136.

126 Significantly, the majority did not consider Harring-

ton to be a case of interlocking confessions. The statementby Harrington was a piece of circumstantial evidence nodifferent from that provided by a witness who couldtestify that Harrington was around, but who could nottestify to any criminal activity.

127 Much of the confusion in Schnebte arises from thevarying views amongst the members of the Court as tothe legality of the confession. Despite the limitation ofthe grant of certiorari to the "confrontation" question,the "confession" split on the Court has an obvious effecton the opinions.

128 Schneble's first "confession" was not reliable. Hissecond "confession" was. The second confession occurredafter what the dissent characterized as "a series of bizarreacts by the police." 405 U.S. at 434 (Marshall, J., dis-senting). The Florida Court found the second confessionwas sufficiently "attenuated" to allow its presentation tothe jury. The jury's handling of the confession, anddetermination of whether it was voluntary, was part ofthe disagreement between the majority and the dissentas to the harmlessness of the offer of the codefendant'sstatement.

Harrington: "Unless the Court intends to emasculateBruton, supra, or to overrule Chapman v. California,supra, sub silentio, then I submit that its decision isclearly wrong."'

2

Which of the precedents, Bruton or Chapman, was

the target in Schneble became clear in Parker v.Randolph.s3° Justice Rehnquist, the author of Schne-ble, writing for a plurality in Parker, resurrectedDelli Paoli and limited Bruton to those cases in whichthe implicated defendant "made no extrajudicialadmission of guilt."' 1 31 The Parker plurality opinion,on confrontation grounds, was inconsistent withBruton, relying instead on the two harmless consti-tutional error cases, Harrington and Schneble. Parkerdid to Bruton what Marshall suggested that Schnebledid to Bruton, assuming Schneble to be a confronta-tion, not a harmless error, case. Justices Stevens,Brennan, and Marshall found Rehnquist's con-frontation position untenable. Justice Powell tookno part in the decision. Justice Blackmun specifi-cally rejected Justice Rehnquist's position on Brutonbut voted with him because he believed that theerror was "harmless."

Justice Blackmun's use of harmless constitu-

tional error in Parker is a classic example of anappellate procedural doctrine blocking the resolu-

tion of an important constitutional question. Hadharmless constitutional error not been available toJustice Blackmun, he probably would have joinedJustice Rehnquist on the merits, providing a ma-jority for the Rehnquist position. While he rejectedRehnquist's Bruton position to make his harmless

error point, 3 2 his harmless error reasoning is diffi-cult to distinguish from Rehnquist's Bruton posi-tion. While one might have substantial disagree-ment with Rehnquist's Parker position, a constitu-tional position ought not fail of a majority for aharmless error doctrine.

Ironically, harmless error is based on a concernfor judicial economy. Had Blackmun prevailed,still another hearing and another round of appealswould have resulted, all in the interest of judicialeconomy.sss The right of confrontation is a consti-

:2

9 Id. at 437 (Marshall, J., dissenting).30 442 U.S. 62 (1979).

131 Id. at 74.132 Id. at 77 (Blackmun, J., concurring in part).3 3Justice Blackmun, concurring in Moore v. Illinois,

434 U.S. 220, 233 (1977), begins to hint at a cure worsethan the disease for the extra round of hearings. Heconsidered the victim's in-court identification of herassailant and said: "[Tihe conclusion that it was harmlessseems to me to be almost inevitable." Id. at 234. Whilehe went on to say that it was for the lower court'sdetermination in the first instance, his apparent willing-

[Vol. 71

HARMLESS ERROR: CONSTITUTIONAL SNEAK THIEF

tutional right. It runs consistently against the gov-ernment and in favor of the defendant. It dependsupon the courts for its definition. Because theharmless constitutional error doctrine exists, andwas able to draw the vote of a Supreme CourtJustice, the status of the right and the case, Bruton,which defined it is still in doubt. If Bruton does notstate the law of confrontation, and arguably it doesnot, ajudicial economy doctrine that does not workought not deny us the knowledge.

When a Supreme Court Justice can frustrate theconstitutional process by refusing to decide a mat-ter on the merits in favor of a procedural doctrineinvented to avoid retrials over omitted "the's," theloss is exceeded only by the danger of that samedoctrine changing the constitutional process with-out warning. The plurality in Parker was not doinganything that the majority in Harrington and Schne-ble had not done under the guise of "harmless"error. The difference is that Harrington and Schneblegave poor notice that the Bruton doctrine would notapply to cases in which there was an extrajudicialstatement from the defendant complaining of theconfrontation failure. The harmless constitutionalerror doctrine is a particularly evil erosive mecha-nism. Assuming for a moment that methods toerode constitutional doctrines without destroyingthem are useful tools in the law, harmless erroranalysis is not one of those useful tools. Harmlesserror analysis does not address, nor give the oppor-tunity to address, the merits of the doctrine beingeroded. The case for gradual change in constitu-tional propositions by erosion is not made by resortto a doctrine that gives proponents and opponentsof the constitutional proposition no opportunity tospeak to the merits and courts no opportunity tosignal where they are going.

Milton v. Wainwright"" is an example of the evileffect of the harmless constitutional error doctrinewhen five Justices are willing to avoid the issue asJustice Blackmun did in Parker. George Milton wasindicted for murder, had a lawyer, and was in jailawaiting trial. He had made some statements thatwere less than fully reliable.sss The police were

ness to reach the conclusion quickly, coupled with ChiefJustice Burger's procedure in Milton v. Wainwright, 407U.S. 371, to consider harmlessness as a new matter in theSupreme Court, suggests the possibility of an increase inharmlessness decisions from the Supreme Court as amethod to decide each case on its narrowest ground.

'34 407 U.S. 371 (1972).135 Chief Justice Burger's opinion does not mention

that the "overwhelming" evidence was a set of "confes-sions" taken over an eighteen-day period, during which

sufficiently concerned that they sent an officer intoMilton's cell, posing as a prisoner. During the twodays that Milton and the policeman, Langford,"roomed" together, Milton made damaging state-ments to the policeman which Langford related tothejury at trial. The District Court viewed Milton'shabeas complaint about the statements as raisingthe issue presented in Massiah v. United States,13 6 butdenied Milton any relief because Massiah had neverbeen held to be retroactive.

Chief Justice Burger, speaking for five membersof the Court in Milton, did not address the Massiahissue.' 37 He assumed arguendo that there was error,and wrote an eight-page opinion explaining whythe hypothetical error was harmless. Justice Stew-art, the author of Massiah, speaking for four mem-bers of the Court, took vigorous exception to thelower court's view of Massiah, the majority's failureto consider Powell v. Alabama,is1 and the Court'sconclusion that the error-whatever it was-washarmless. One hardly needs to take sides on themerits of the Massiah retroactivity issue to demon-strate that the harmless error doctrine is an abom-ination when constitutional matters are at stake inthe Supreme Court. The retroactivity of Massiahremains unanswered and the continued strength ofPowell is now in doubt because five out of ninemembers of the Court were able to classify anunidentified error as harmless beyond a reasonabledoubt. Laying aside the value of five out of ninepersons finding anything true beyond a reasonable

the defendant was held incommunicado and questionedalmost every day, often for hours at a time. He deniedhis guilt for the first ten days. See 407 U.S. at 383. ChiefJustice Burger gave no consideration to the possibilitythat the jury may well have been unwilling to convictbased upon the suspicious confessions, and gave themgreat weight only in view of the "harmless error"-the"voluntary" conversation with the cellmate.

'36 377 U.S. 201 (1964). United States v. Henry, 100 S.Ct. 2183 (1980), decided eight years after Milton, pre-sented almost exactly the same Massiah-relevant facts.Chief Justice Burger's majority opinion in Henry detailedan understanding of Massiah with which four Justicestook issue on one basis or another. Nothing in the ChiefJustice's opinion justifies, explains, or even mentions theeight years in which the Court left everybody else up inthe air as to the meaning of Massiah.

137Justice Stewart, the author of Massiah, described itas a "counsel" case in Milton, while the-district and circuitcourts read it to be about "voluntariness." 407 U.S. at380 (StewartJ, dissenting). The ChiefJustice's excursioninto harmless constitutional error as a matter of firstinstance in the Supreme Court left the disagreement,along with the retroactivity question, unresolved.

' 287 U.S. 45 (1932).

STEVEN H. GOLDBERG

doubt, substantial questions remain unansweredand new ones are raised unnecessarily. In addition,the Supreme Court spends its valuable time mak-ing a factual judgment about a long record ratherthan examining the law of the case, or, at least, thelaw of some other case. The doctrine of harmless

error damages the Constitution by failing to ad-dress important issues as in Milton. It also causes amisallocation of court time which could be better

spent performing tasks other than reading a longrecord. The harmless constitutional error doctrinein the Supreme Court is an unwarranted and

dysfunctional impediment to the constitutionalprocess.

The constitutional damage from the harmlessconstitutional error doctrine is not limited to theSupreme Court and its use of the doctrine to avoidconstitutional judgments as in Parker, or to makeconstitutional judgments without real notice as inSchneble. Appellate courts, both state and federal,have occasionally been less than enthusiastic aboutcertain constitutional decisions of the United States

Supreme Court. Harmless constitutional error pro-vides such courts an opportunity to "destroy ordilute constitutional guarantees."1 39 The doctrineis unmatched as a tool for the secret theft ofconstitutional rights. A finding that a constitu-tional error is harmless is almost beyond questionor review. While there are some fact circumstancesand trial situations so obvious that it is incrediblethat any appellate judge would make a harmlesserror finding,4 ° most such findings are based uponlong and relatively inaccessible records. Further,

the majority opinion will generally reflect thoseportions of the record which support the "harm-less" determination. If there is no dissent on the

139Justice Harlan's Chapman dissent recognized thedanger and reserved the eventuality of state court dilu-tion of federal constitutional rights as a basis for hisconcurrence with a decision which would prohibit theactivity by imposition of a federal standard. 386 U.S. at50.

140 A rare example of a harmless constitutional errordecision which is so preposterous on its face that noparticular resort to a record is required is found in Allisonv. Gray, 603 F.2d 633 (7th Cir. 1979). The error was afailure to allow the defendant to present his alibi evi-dence. The Wisconsin Supreme Court and the federaldistrict court apparently believed that the government'scase was so overwhelming that there was no need for ajury to listen to and believe the defendant's case. Short ofthis kind of a case, in which the court finds the failure ofthe government to allow a trial to be harmless, harmless-ness is not easily seen by the reviewer without benefit ofa full record.

law, there is not likely to be a dissent on the facts-even if a judge might have some question aboutthe facts. 4'

If a state or federal appellate court chooses touse the harmless constitutional error as a tool toemasculate a constitutional right through a con-sistent finding of harmlessness in a series of cases,it is likely to succeed without question. Althoughprosecutors who handle cases within a jurisdictionmay watch closely to see which area of error isbecoming "harmless,' ' 142 commentators, the peoplemost likely to be concerned with a pattern ofdilution, are unlikely to be alerted, given the lickof legal mistake with which to quarrel. A givendefendant may press through the habeas corpusmaze to appeal a state determination of harmless-ness or appeal to the Supreme Court from a circuitcourt harmlessness decision, but the likelihood of asuccessful appeal of a harmlessness affirmance isslim. Given the number of harmless constitutionalerror decisions, and the difficulty in finding rever-sals of those decisions by a higher appellate court,the potential of the harmless constitutional errordoctrine for unmaking law has not gone unnoticedby appellate tribunals. 43 While there may be someother explanation for the marked increase in theuse of harmless constitutional error, avoidance of

141 Milton v. Wainwright, 407 U.S. 371, exemplifiesthe situation where there is a quarrel about the factualfinding of harmlessness which would not likely haveoccurred but for the legal dispute. Even in Harrington andSchneble, the dispute on the facts is only in furtherance ofthe complaint about the test the Court is adopting.

142 The problem of prosecutorial deterrence is consid-ered at text accompanying notes 154-159 infra. JusticeCameron of the Arizona Supreme Court has observedthat the result of consistent signals to prosecutors isprosecutorial reception of the signals: "[I]f a particularerror is declared to be harmless a sufficient number oftimes, then the cumulative effect of such holdings will bethat both the prosecution and the trial judge will tend toignore the error and commit it again." Cameron & Duke,When Harmless Error Isn't Harmless, 1971 LAw & Soc. ORD.23, 42.

143 Whether the absence -or presence of second levelreversals of harmless determinations would be the moreunhappy circumstance is unclear. Given the large num-ber of harmless cases, it seems fair to conclude that thesmall number of second level reversals is directly relatedto the difficulty of review of a factual judgment. On theother hand, a significant number of second level reversalson harmless error would be sorry proof that lower appel-late courts were, indeed, hiding behind harmless error asa means to dilute constitutional rights. The other possi-bility, that there is an incredible increase in the numberof trial mistakes that do not matter in fact, has little tosupport it in logic or evidence.

[Vol. 71

HARMLESS ERROR: CONSTITUTIONAL SNEAK THIEF

constitutional responsibility is not without prece-dent. 1"

Harmless constitutional error presents specialproblems for trial courts and prosecutors that arenot a concern when the harmless error is "com-mon." Because harmless constitutional error alwaysoperates in favor of the government and againstthe defendant, the doctrine puts unique pressureon a conscientious trial judge or prosecutor toignore a citizen's acknowledged constitutionalrights based upon the "no harm-no foul" the-ory.1

4 5

Trial courts make the first, and often most influ-ential, decision concerning the constitutionalrights of the defendant. It is the trial court, andparticularly the state trial court 147 where the cost

144Justice Powell, concurring in Argersinger v. Ham-lin, 407 U.S. 25, 65 (1972), had no hesitancy in ascribing"the failure of many state courts to live up to theirresponsibilities in determining on a case-by-case basiswhether counsel should be appointed" as a motive behindthe landmark decision in Gideon v. Wainwright, 372 U.S.335 (1963).

145 Although demonstrating the effect of any decision-maker's benign neglect is difficult, the art form fromwhich the phrase is taken, professional basketball, pre-sents an interesting analogy. Before the "no harm-nofoul" type interpretations of the rules by modern referees,basketball was considered a kind of sporting dance in-appropriate (in those days) for "men." Anyone watchingthe modem no harm-no foul game can attest that aportion of that contact sport resembles a war zone. It isdifficult for one trained in the law to assume that prose-cutors and judges would not be at least as quick asbasketball centers and referees to adjust to this kind ofdevelopment.

146 The first judgment is the standard against whichother possibilities are considered. The trial court's judg-ment about both the facts and the law often influencesdecisions three or four appellate layers later. See, e.g.,Rhode Island v. Innis, 100 S. Ct. 1682, 1690 n.9 (1980)(quoting trial judge), where the trial judge's statementthat it was "entirely understandable that [the officers inthe police vehicle] would voice their concern" made it allthe way to a United States Supreme Court opinion andserved as a part of the foundation for the proposition thatthere was no "interrogation." See also Note, Fifth Amend-ment-The Meaning of Interrogation Under Miranda, 71 J.CRIM. L. & C. 466 (1980).

147Justice Harlan consistently maintained-that thecircumstances of the states were very different from thecircumstances of the federal government. His argumentin Chapman, that federalism should be the first constitu-tional principle, persuaded no one on the Court. Regard-less of the impotence of the argument for law, the fact isthat there is a huge difference between the circumstancesand relationships of a federal trial court and a state trialcourt. The discussion which follows assumes a trial in astate trial court, the forum in which the vast majority ofcriminal litigation takes place.

of the exclusion of evidence, for instance, is mostimmediately appreciated. Justice Frankfurter's ad-monition that "the safeguards of liberty have fre-quently been forged in controversies involving notvery nice people,"'" is rarely lost on a trial judge,who is usually looking over the bench at the "notvery nice" person. Unfortunately, there is nomethod for documenting the effect of the harmlessconstituti6nal error doctrine on trial courts. Asidefrom the fact that most trial court opinions gounpublished, the issue of "harmlessness" is neverofficially raised at the initial confrontation over theloss of a constitutional guarantee.

Some observations about the operation of trialcourts and their place in the community mighthelp fill the void left by the failure of formalopinions. Acknowledging the danger in anecdotalproof, I doubt that I am the only lawyer in theland who, when citing a precedent for error, hashad the trial court review the precedent and say:"Counsel, I agree that the court held that to beerror. But they said that it was harmless, and Ithink that this case is virtually the same as yourprecedent."' 49 Predictably, the trial court focuseson the precedent's harmlessness determinationrather than on the determination that there wasindeed an error.

Most constitutional matters arise in special pre-trial hearings under the glare of some publicity."50

Significantly, appellate courts decide such mattersfar away and after the heat is off. Trial judges livein, socialize with, and are, by design, subject to theinfluence of the community.' 5' What ought such a

14' United States v. Rabinowitz, 339 U.S. 56, 69 (1950)(Frankfurter, J., dissenting), overruled, Chimel v. Califor-nia, 395 U.S. 752 (1968).

149 This composite response reflects some experience inresponse from trial judges commenting upon precedentwhich had good news and bad. The good news was theerror, the bad news was that it was harmless. While acareful advocate would find precedents which showedonly the error, the opportunity for that selection de-creased rapidly after Chapman as more and more casesfound both error and harmlessness.

150 Gannett Co. v. DePasquale, 443 U.S. 368 (1979),demonstrates that the pretrial hearing is often of as muchpublic interest and concern as the trial. The trial court'sdecision to close the hearing may reflect something morethan a concern for an impartial jury. The SupremeCourt's willingness to allow the closure may evidence anunstated understanding of the pretrial pressures on judi-cial decisionmakers. See also Richmond Newspapers, Inc.v. Virginia, 100 S. Ct. 2814 (1980) (trial must be open tothe press); Note, Constitutional Right of Access to CriminalTrials, 71 J. CRIM. L. & C. 547 (1980).

' Even in those "enlightened" jurisdictions which

STEVEN H. GOLDBERG

judge do if a defendant charged with murder, andagainst whom there is overwhelming evidence ofguilt, makes a motion to suppress evidence becauseof a violation of the Constitution? Justice Cardozo'sreminder 52 of the high price society pays for adecision to exclude evidence is best understood bya trial judge. To that problem for the trial court,add the proposition that appellate courts may findconstitutional error to be "harmless" and that in arelatively similar case the highest court of the statehas once done just that. The trial court can readthe highest court of the state to be holding that theerror in question will always be harmless. 53 It mayread the court to be signaling a change in the law,and it may be correct. Last, and for constitutionalprotection, probably worst, the trial court may becorrect in its view that the instant case and theprecedent are factually so similar that the sameharmless constitutional error determination wouldprobably obtain upon review. The result will bethe same, regardless of which view the trial courttakes. Either because the error is always harmless,or the law is changing, or because it can accuratelypredict that this particular error in this context willbe harmless, the trial court will consciously commitconstitutional error, safe in the "instruction" to doso from the state's highest court. The result is adoctrine which, in the guise of judicial economy,almost requires the trial court to initially ignorethe enforcement of constitutional rights. Signifi-cantly, the hypothesized activity does not depend

have adopted plans for judicial selection other than theelection contest, some form of community veto existsupon the performance of judicial trial function. Mostjurisdictions maintain the election of trial judges, andwhile the upset of a sitting judge is rare, it does happen.More importantly, the trial judge, as opposed to theappellate judge, suffers or enjoys all of the personalconcerns for acceptance within the community of whichthe judge is not only an integral, but an important andvisible part.

152 See note 116 & accompanying text supra.153 At least one court has determined that a constitu-

tional error can never be harmful. In State ex rel. Me-Mannis v. Mohn, 254 S.E.2d 805 (W. Va. 1979), theWest Virginia Supreme Court of Appeals held that viola-tion of the principle in Estelle v. Williams, 425 U.S. 501(1976) (defendants may not be forced to attend trial inprison garb), was harmless constitutional error when thedefendant was on trial for a crime committed while inprison, and when the defense failed to raise its objectionuntil after calling its first witness. Whether the decisionmight have been justifiable under a due process theory,or whether Estelle is distinguishable (the court's opinion)was beside the point. The West Virginia Court chose tocreate the anomaly that certain activity was always errorand always harmless.

upon any ulterior motive upon the part of the trialcourt. Quite the contrary is true. A respectableargument can be made that when a local trialjudge is faced with the possibility of turning a"murderer" loose for the violation of a constitu-tional right which an appellate court has said inthe same context led to a harmless constitutionalerror, the judge should err in favor of communitysafety. A "correct" result of guilty, given the cer-tainty, likelihood, or possibility that the appellatecourt will say that the trial court's error was harm-less is more appropriate than any other result. It isalso easier, and in this situation "easier" is notnecessarily pejorative.

The result of the harmless constitutional errordoctrine at the trial court level is to encourage thediminution of rights against the government for allindividuals against whom the state has an over-whelming case.' 54 Harmless constitutional errorthen becomes the presumption rather than theexception. The fault is not with the trial court butwith the concept that the abrogation of a consti-tutional right that runs consistently against thegovernment can be harmless.

As difficult as it is to isolate the effect of theharmless constitutional error on trial courts, it iseven more obscure with prosecutors. There is, nev-ertheless, the nagging suspicion that the majordenial of constitutional guarantees resulting fromthe harmless constitutional error doctrine relates toits effect upon prosecutors. Again, one need notposit the evil-minded prosecutor to demonstratethat harmless constitutional error steals significantrights by its effect upon the prosecuting authority.Prosecutors, as with all lawyers, are trained torepresent their client to the limit of the law. Theprofession considers it unethical and unthinkablefor the advocate to cross over the line, and mal-practice not to approach that line as closely aspossible in pursuit of the client's just cause. Theprosecutor's duty to "justice,' ' 15 may raise some

154 The difference between the "effect of the evidence"test and the "overwhelming evidence" test may be ofsignificance in this situation. At least with the formertest, the trial court is not encouraged to prejudge thestrength of the prosecution's case and make an error/noerror decision based upon that judicial prejudgment ofguilt.

'S A sense of uneasiness pervades the common sugges-tion that prosecutor/advocates are somehow differentfrom other advocates and can psychologically suffer pull-ing selected punches in the interest of justice. See, e.g.,Standards Relating to the Prosecution and the Defense Function,ABA STANDARDS RELATING TO THE ADMINISTRATION OF

CRIMINALJUSTICE, Standard 3-1.1 (approved draft, 1979)

[Vol. 71

HARMLESS ERROR" CONSTITUTIONAL SNEAK THIEF

doubt about how he should respond to a constitu-

tional violation, but the prosecutor's instincts as alawyer combined with the harmless constitutional

error doctrine, demand that the prosecutor abdi-

cate any role as a positive force for the maintenanceof constitutional guarantees.15

When constitutional rights are at issue in the

courts, prosecutors have a consistent position.

Either as the result of the facts they are given or as

a matter of government or personal policy, the

prosecutor is arguing in court that an alleged vi-olation of a constitutional right is not a violation,

or if it is, should not vitiate a good case. 157 Thesame is not necessarily true when nonconstitutionalrules are at stake. The result is that the prosecutor

contributes to the erosion of doctrines in the courts

by maintaining a consistent position with respectto the harmlessness of constitutional error. To the

extent that any litigant can find the opportunityor mechanism to argue harmlessness to the appel-late tribunal, it is the prosecutor. Furthermore,

since the prosecutor is a consistent litigant against

opponents with one-case or one-issue interests, theprosecutor can, over a series of cases, maintain a

position which is beyond the power of any singleopponent to effectively confront.

The harmless constitutional error doctrine even

affects the prosecutor in the investigation and trial

stages. When an advocate is faced with a trialburden of proof beyond a reasonable doubt, he has

the inclination and opportunity to watch the casesto see which errors are consistently harmless andwhich are not. Where the trial task and individual

("The duty of the prosecutor is to seek justice, not merelyto convict."). It is difficult to include under that umbrellathe suggestion that a prosecutor is somehow bound notto take advantage of evidence which that prosecutorknows will be considered "harmless" even if it might betechnical constitutional error.

"56 The prosecutor's potential as a force for the pres-ervation of constitutional values is twofold. While theprosecutor has an obvious role in refusing to presentevidence which will result in reversible error, in manyinstances the prosecutor also can be effective in directinglaw enforcement policies and activities.

157 Any suggestion that the office of prosecutor presentsa unique circumstance in which the office holder has anobligation to suppress constitutionally infirm evidence orat least refuse to argue its harmlessness is based on amisunderstanding of the adversary system in fact, andprobably in theory. See, e.g., Milton "v. Wainwright, 407U.S. 371, in which at least one prosecutor argued "harm-lessness first" all the way to the Supreme Court. It is theprosecutor's continuing stake in the merits and ability toprofit from consistent "harmless" decisions in lieu of avictory on the merits that helps to make the doctrine ofharmless constitutional error particularly pernicious.

temperament demand every available weapon, the

doctrine presents a situation in which only themost foolish of prosecutors would avoid the risk of

using the error. Every time an error is declaredharmless in a particular situation, it diminishes the

risk to the prosecutor in the use of the evidence or

the technique.1 a The lessening of the risk is addedinto a formula which favors risk-taking based uponthe doctrine alone. In a sense, the doctrine encour-

ages the prosecutor to use the evidence or the

technique in every case. Initially, there are three

15 Situations in which prosecutorial conduct may be

influenced by "harmless" decisions are difficult to iden-tify'. It may be reckless to ascribe motive from a consistentline of cases, but a series of decisions in the MinnesotaStipreme Court leaves the distinct impression that pros-ecutors respond to what they read.

In State v. Jones, 277 Minn. 174, 152 N.W.2d 67(1967), the Minnesota Supreme Court reversed a convic-tion because of prosecutorial misconduct. Among otheritems the prosecutor argued to the jury the failure of awitness to answer certain questions. Id. at 183, 152N.W.2d at 74. The reversal was, however, the result ofmultiple sins. State v. Russell, 282 Minn. 223, 164N.W.2d 65, cert. denied, 396 U.S. 850 (1969), concernedan improper argument wherein the prosecution alludedto the defendant's failure to call witnesses. The courtsaid, "we do not approve," but refused to reverse becausethe remainder of the evidence was so overwhelming. Id.at 228, 164 N.W.2d at 68. In each of the next seven years,the court had at least one occasion to tell prosecutors thatit was error to comment upon a defendant's failure tocall witnesses, and in no instance did it reverse for thatreason. Most of the cases come from the same prosecutor'soffice. All but one came from the three major metropol-itan prosecutors' offices. While it cannot be shown thatthe continued comment on the failure to call witnesses isa direct result of the failure of the court to reverse, thecourt kept affirming and the prosecutors kept comment-ing. See State v. Spencer, 311 Minn. 222, 248 N.W.2d 915(1976) (defendant's testimony); State v. Redd, 310 Minn.145, 245 N.W.2d 257 (1976); State v. Fields, 306 Minn.521, 237 N.W.2d 634 (1976); State v. Meadows, 303Minn. 76, 226 N.W.2d 303 (1975); State v. Caron, 300Minn. 123, 218 N.W.2d 197 (1974); State v. White, 295Minn. 217, 203 N.W.2d 852 (1973); State v. Bell, 294Minn. 189, 199 N.W.2d 169 (1972).

Two related lines of cases show the same tendency for"repeat business," or prosecutorial conduct which is "dis-approved" but for which the Court continuously refusesto reverse. See generally State v. Shupe, 293 Minn. 395,196 N.W.2d 127 (1972). Shupe disapproved the prosecu-tor's implication to the jury that he would have providedmore evidence but for some circumstance. In at leastthree subsequent cases the same comment was disap-proved without reversal. In State v. Thomas, 307 Minn.229, 239 N.W.2d 455 (1976), the court acknowledgedthat its admonition to prosecutors in a 1933 case regard-ing the misrepresentation of the presumption of inno-cence was going unheeded. It did not reverse but said itwould henceforth. Eight cases later it had not reversedand gave no indication that it ever would.

STEVEN H. GOLDBERG

possibilities: (1) the evidence or technique does notinvolve any error, (2) if the evidence or techniqueinvolves error, it will be harmless, and (3) theevidence or technique involves error that will causea reversal because the remainder of the evidence isnot "overwhelming." What should the intelligentand conscientious risk assessor do? The first twopossibilities present no question. If there is no errorthere is no problem, and if the error is harmless theonly problem is the time and expense of an appeal.The result is the same: a legal conviction. Convic-tions which are legal are, after all, what the societypays the prosecutor to obtain. The third choice isthe problem. The court has defined a doctrine ofharmless constitutional error which says to a pros-ecutor that if the case is not overwhelming anyerror will cause a reversal, and if it is overwhelming,no worry. The prosecutor then looks at the caseand determines that it is not very strong. Use ofthe evidence or technique has two chances of suc-cess-no error and harmless-and one chance offailure. By the Court's definition of "harmless,"that one chance of failure demands that the evi-dence or technique be crucial to the prosecutor'scase. The prosecutor has no advocate's choicewhich mitigates in favor of not using the evidenceor technique. Even if the prosecutor believes thecase is strong, the likelihood is that the evidence ortechnique will be used. The odds are still two-to-one. Further, the advocate's predilection to coverevery base is reinforced by the doctrine's admoni-tion: if the evidence or technique is not needed bythe advocate it is not likely to cause a reversal. 59

The effect of the rule upon the government'sadvocate is particularly significant. Prosecutors areamong the most deterrable of groups. If any of theSupreme Court decisions of the last quarter centuryhave a deterrent effect, prosecutors are the primegroup to respond. Not only are they affected di-rectly by reversals, they are sufficiently trained andhave sufficient time to fully appreciate the prece-

1s Harrington v. California, 395 U.S. 250 (1969), was,according to Justice Douglas, just such a case. His view,represented by Justice Marshall's dissent in Schneble v.Florida, 405 U.S. at 433, was that the Harrington evi-dence was merely cumulative. The confession in Miltonv. Wainwright, 407 U.S. 371, seems another example ofjust one more piece of evidence which a prosecutor placedupon an already very large pile. The best proof thatprosecutors are likely to use all they have, even if theydon't need it, is the number of times that appellate courts,by finding harmless constitutional error, have said thatthe prosecutor used evidence that wasn't needed.

dent.' 6° The harmless possibility, then, tempts agroup that might in its own self interest be asignificant force for preserving the constitutionalright and, in addition, influence law enforcementto do the same.

V. No RHYME NOR REASON

The prosecutor's position with respect to a con-stitutional right which may lead to error exempli-fies the last, and most ironic, problem with theharmless constitutional error doctrine. Harmlesserror exists only as a means of judicial economy,and harmless constitutional error does not eveneconomize judicial time or activity.1

61 The judicial

economy theory assumes that a rule which allowsreversal only when the error is substantial willdiminish the number of retrials and thus cut judi-cial time and speed matters to conclusion.' 62

The harmless constitutional error doctrine, as itaffects the prosecutor, contributes to court conges-tion. Although quantification is impossible, thedoctrine clearly causes some diminution in theprosecutor's concern that abrogation of a consti-tutional right will cause a reversal. Consequently,the prosecutor will use the evidence or techniquein a number of cases which, without the doctrine,he would not. Each such case represents one trialwhich, without the doctrine, might not have takenplace. In the event of a conviction, an appeal isalmost guaranteed because there is an obviousconstitutional error. If the prosecutor was right andthe error was, indeed, harmless, the net effect ofthe use of the evidence under the shield of thedoctrine is one more appeal. If, on the other hand,the prosecutor's guess was wrong, and the errorwas harmful, there is one more appeal and onemore trial, where, without the doctrine there might

'60 One of the major complaints about the extension ofthe exclusionary rule has been that policemen have nei-ther the training to understand it nor do they findthemselves in situations in which their action can beaffected by an arcane judicial precedent. See Bivens v. SixUnknown Named Agents, 403 U.S. 388 (1971). Prosecu-tors, on the other hand, are the specific group at whichexclusionary deterrence might properly be aimed.

161 Empirical proof of economy is virtually impossible.Professor Saltzburg, supra note 4, at 1032 n. 158, presentsan interesting view of the de minimus effect of harmlesserror in criminal trials in the federal system. Nothing ineither the cases or the literature presents a detailedargument for the proposition that harmless constitutionalerror presents the same judicial economy claimed forharmless common error.

'6 See generally State v. Link, 289 N.W.2d 102 (Minn.1979).

[Vol. 71

HARMLESS ERROR: CONSTITUTIONAL SNEAK THIEF

have been no trial in the first instance. There is noinstance in which the harmless constitutional errordoctrine encourages prosecutors not to file caseswhich would not otherwise be tried, and manychances for the filing of additional cases becausethe doctrine exists.

The effect of the doctrine on trial courts alsocontributes to court congestion and works againstthe rationale that supposedly supports the doctrine.To the extent that the doctrine encourages a trialjudge to allow evidence even though it is likelyerror to do so, the doctrine destroys another oppor-tunity to diminish litigation. In some cases, if thetrial court suppressed the evidence, the prosecutorwould decline to proceed. To whatever extent thedoctrine discourages the suppression, it increasesthe number of trials. In addition, each time thefailure to suppress involves a constitutional error,it almost guarantees an appeal and another fullrecord hearing in the appellate court before thelitigation is laid to rest because the error washarmless.

Moreover, it is not clear that the harmless con-stitutional error doctrine contributes to judicialeconomy at the appellate level. Undoubtedly, somecases are not retried because the court found errorbut found it harmless. How large that numbermight be is as difficult to ascertain as some of theother possibilities which have been discussed. It is,however, fair to observe that in order for the errorto be harmless the case against the appellant mustbe "overwhelming" without the error. If the gov-ernment's case is that good-and the defendanthas already seen it succeed once-the chance for aplea after the appeal is enhanced. If the court findsthe error and does not have the harmless errordoctrine to fall back on, the government is moreinterested in the defendant's plea than it was beforetrial. Whatever the number of retrials avoided bythe harmless constitutional error doctrine, it isdiminished by the increased likelihood of a plea ifthe doctrine is abandoned.

Importantly, when the error in question is con-stitutional rather than common, any resultant trialeconomy is offset by the certainty of the increase inappeals. While a defense attorney may persuade adefendant that a truly insignificant common erroris not an adequate foundation for an appeal, thatsame result is not likely when both the lawyer andthe defendant know the Constitution is implicated.The net economy effect of the doctrine upon ap-pellate courts and trial courts cannot be left with-out noting that the issue of harmlessness raises yet

one more determination from which the defendantmight take yet one more appeal. For every trialthat the harmless determination might save, thereis an appeal to be spent."s

A last unique aspect to constitutional error dif-ferentiates it from common error when the judicialeconomy argument is made for a harmless errordoctrine. The major reason for including the crim-inal trial within the original harmless error rulewas the prevailing practice among defense attor-neys of putting error in the record as a hedgeagainst a guilty verdict.' 64 The effect of that system,which provided the only litigant with an interestin never reaching a decision a technique to foreverforestall the reaching of a decision, was to neverreach a decision. Constitutional error, however, isbeyond the control of the defense lawyer for themost part. It is difficult to place a bad search, abad statement, a bad lineup, the failure of counsel,or other government error into the record. Theneed for a harmless error rule to avoid retrialscaused by defendants does not exist for constitu-tional error.

VI. CONCLUSION

The harmless constitutional error doctrine sharesneither history nor logic with the harmless errordoctrine from which it was wrenched in Chapman.It offers neither logic nor method for finding prooffor the proposition that judicial economy resultsfrom the operation of the doctrine. Indeed, thedoctrine arguably contributes to the amount ofjudicial time spent on criminal cases. Unfortu-nately, the doctrine enjoys great popularity withappellate courts as a mechanism for reducing thestrict enforcement of criminal constitutional pro-cedure rules.

The effect of the doctrine upon precedents defin-ing constitutional criminal procedure-their crea-tion, maintenance, and change-is devastating. Asto constitutional rights there should be no harmlesserror.165 The Court should adopt a rule of auto-

'63 Constitutional error presents the added possibilityof federal habeas corpus. But see Stone v. Powell, 428 U.S.465 (1976) (ending federal habeas corpus in fourthamendment cases). Insofar as many jurisdictions havetwo levels of appellate courts, and the federal habeasprocedure might involve three levels of review, a strongargument can be made that it takes more time going upthe ladder to resolve the harmlessness issue than it wouldtake to go back down and try the case without the error.

164 See note 14 supra.'65 ChiefJustice Burger concedes that thejudicial econ-

omy realized by saving one jury trial is outweighed by

19801

STEVEN H. GOLDBERG

matic reversal, fulfill its function with respect tothe Constitution, and make its judgments in fulllight of the undiluted effect of the rules it makes.

A terrible symbolic price is paid for maintenanceof a harmless constitutional error rule. There is avisceral, if nonlegal distaste for the proposition thatan individual's loss of a constitutional guarantee,protected only by the courts, can somehow be"harmless." An improper search of a home shouldnot be equated with a state's omission of the word"the" from the defendant's charging papers. Law-

the interest in safeguarding the rights of the individualdefendant. Writing for a unanimous court in Standeferv. United States, 100 S. Ct. 1999 (1980), the ChiefJusticeheld that the civil doctrine of nonmutual estoppel wouldnot be applied in criminal cases because the "importantfederal interest in the enforcement of the criminal law"outweighed the "economy concerns that undergird theestoppel doctrine." Id. at 2008. In distinguishing thecriminal from the civil, he concluded with an approvingquotation of the Court of Appeals rationale for thatdistinction:

"The purpose of a criminal court is not to provide aforum for the ascertainment of private rights. Rather

yers should pause at the proposition that govern-ment can violate a basic restriction upon itself and,through a court, tell the individual who was thebeneficiary of the restriction: "no harm-no foul. ' i66

There is something disquieting about the admis-sion that constitutional rights are so often abro-gated that we need a separate doctrine to excusesome of them so that our decisionmaking systemwill not break down. Finally, there is some shamein trying to explain how the loss of a constitutionalguarantee is harmless beyond a reasonable doubt-five-to-four.

it is to vindicate the public interest in the enforce-ment of the criminal law while at the same timesafeguarding the rights of the individual defendant.The public interest in the accuracy and justice ofcriminal results is greater than the concern forjudi-cial economy...."

Id. (emphasis added) The judicial economy concern inboth the nonmutual estoppel situation and in the harm-less constitutional error case is exactly the same: one jurytrial.

166 "No man ought certainly to be a judge in his owncause, or in any cause in respect to which he has the leastinterest or bias." THE FEDERALIST No. 80 (A. Hamilton).

[Vol. 71