University of Michigan Law School University of Michigan Law School Scholarship Repository

Articles Faculty Scholarship

1999 Lost Lives: Miscarriages of Justice in Capital Cases Samuel R. Gross University of Michigan Law School, [email protected]

Available at: https://repository.law.umich.edu/articles/637

Follow this and additional works at: https://repository.law.umich.edu/articles Part of the Courts Commons, Criminal Law Commons, Criminal Procedure Commons, and the Evidence Commons

Recommended Citation Gross, Samuel R. "Lost Lives: Miscarriages of Justice in Capital Cases." Law Quad. Notes 42, no. 1 (1999): 83-94. (Based on a paper delivered at the National Conference on Wrongful Convictions and the Death Penalty at Northwestern University Law School in November 1998 and subsequently published under the same title in Law & Contemp. Probs. 61, no. 4 (1998): 125-52.)

This Article is brought to you for free and open access by the Faculty Scholarship at University of Michigan Law School Scholarship Repository. It has been accepted for inclusion in Articles by an authorized administrator of University of Michigan Law School Scholarship Repository. For more information, please contact [email protected].

In case after case, erroneous conviction spared, thus lessening the chance of for capital murder has been provelz. I executing the innocent." contend that tlzese are not disconnected In other words, we need not \vony accidents, but systematic consequences of about this problem because we have the nature of lzomicide prosecution in already taken care of it. The following article is based on a paper deliv- general and capital prosecuLion in How convincing are these three ered at the National Conference on Wrongful particular - Lllat ln this respect, as in premnises? T1ze strong version of the first - Convictions and the Death Penalty, held at others, death distorts and undermines the Judge Halidk position that convictioizs of Northwestern University Law School in the law. innocent people just don't happen - is There are three factual claims behind false. In 1932, Edwin Borchard responded Novembei: At deadline time the complete ver- the that capital convictions of to the clalm that "iimocen~men are never sion was in press for 61 Law & Contemporary innocent defendants are very rare. convicted" by publishing his now classic problems (1998). (1) Erroneous convictions are rare in book, Comiicti~zgTlze bzlzocent, in which criminal rosec cut ions of aizv sort, and their he documented 65 of rlzese cases that danger is greatly exaggerated. Judge never happen. Since then, several other Learned Hand captured tlzis sentiment in compilations of proven erroneous his frequently quoted observation: "Under cornictions have been published, and new our criminal procedure the accused has cases continue to surface svit1-i regularity every advantage. . . . He is immune from Nobody knows the true number of question or comment on liis silence; he miscaken convictions. Since 1992 at least cannot be convicted when there is the least 53 defendants - mostly convicted rapists fair doubt in the minds of any one of the - have been exonerated by DNA twelve. . . . Our dangers do not lie in too identification evidence; most of them were little tenderness to the accused. Our released after spending years in prison. procedure has been alxvays haunted by the These were flukes. The technology to prove ghost of the innocent man convicted. It is their innocence happened to become an unreal dream. What we need to fear is available before the physical evidence from the archaic formalism and the water>? the crime (semen or blood) was lost or sentiment that obstructs, delays, and destroyed, or deteriorated beyond use. Itk defeats the prosecution of crime." anybody's guess how many other innocent (2) On the whole, homicides are easier prisoners haven't had the benefit of this to solve than most other violent felonies. sort of luck. The erroneous convictions that Homicide is typically a crime of passion are discovered may truly be the tip ol an ra~lzerthan design, and the killer is usually iceberg. a relative, friend or acquaintance of the Still, the vast majority ol convicted victim. For example, in 1994, about 78 defendants are no doubt guilty; the iceberg percent oi robberies and 52 percent of - whatever its size - floa~sin a sea of aggravated assaul~sin the United States factually correct decrsions. Learned Hand's were committed by strangers, compared view is simply an example of a common with only about 25 percent of homicides. huinarz tendency to assimilate "usually" to As a result, most l?ormcides present no real "always," and "raroly" to "never." This can question about tlze ideiztily of the criminal, be dangerous. Airplane crashes (or. to and no real nsk of misrake. continue a conceit. collisions bet~rieen (3) Homicides, and capital homicides in ocean liners and icebergs) are also rare; as particular, get far more aLtenLion ban other passengers, we can lee1 cornfortable telling crimes. This suggesls that errors xvill be less ourselves and each other not to worn that likely in these cases because thej~are it wmll never happen. But engineers, uabc examined rvit1-1much more care than controllers and pilots must no1 ignore others. For example, Franli Carrington crashes. These are ten-ible, tragic events, wrote in 1978 " [Olur legal system and they remain rare precisely because as a exarnines capital convictions with such an society we do u1on-y abou~~lzern. and tr)- to incense scrutiny that . . . when there is the stop them from ever happening. slightes~doubt ol guilt (even after The second point - that in rnost colzviction), a comlnuiation will usuallji homicides there is no serious iac~ual resul~,or tlze individual \\ill otherwise be questioiz about the guilt ol [lie accused - I am concerned with any wrongful conviction of a defendant charged with a capital crime, regardless of the crime or the penalty. The worst mistake, the execution of an innocent defendant, appears to be the rarest. This is what we ought to expect: Guilty or innocent, few of those who are sentenced to death in America are actually executed.

1s tme. That reduces the field considei-abl). one or several respects. The case may the demand lor the clea~1-1penalty itself. UnSortunately the ease aitli which most relatively litlle publicity; the murder Some factors nv.1~appear in both groups. hoinicides may he solved does relatively inay be relatively low on the scale of For example, a capital case is lilzely to be little to increase [lie accuracy oi clecision- horror; the clelendant may plead guilty the sort of case that would be highly making in capital homicide cases, since rather khan go to trial, in \vhicli case he publicized in any event, and asking Sou the that subset is likely to include 111ost of [he ~villnormally be sentenced to life deatli penalty is likely to malze it inoi-e so. cases in which factual cletenninations are imprisonment or a term ol years; if he does If capital cases do produce erroizeous most difficult. In most lio~llicidesthe klller go to ti-ial he may be con~ilctedof a non- convictions, there are clifiereiit implications was known to the ~ict~~~l;that is tlie main capital crime, or acquitted altogether; if he depending on the cause ol the erroneous fact that makes most homicides eaS>T to is convicted of a capital crime, he may be conviction. The causes in the first group sol17e.But not capital murders. For selltenced to lile impi-isonmei~t;and finally imply that we should be wary ol iinposin~, example: a study ol honlicicle prosecutions if he is sentenced to death, he will or esecuting death sentences, because lrom 1976 through 1980 in Georgia, probably never be executed. capital cases are of the sort where Floncla, and Illinois found that while only I am concerned kvit11 any w-ongf~11 el?-oneous coiivictions are particularly 17 percent to 22 percent of all the conviction of a defendant charged mith a lilzely regardless of the sanction req~~ested homicide ~lcti~sin those states \arere liilled cap~lalclime, regardless ol he crime or the or imposed. Abolishing the deatli penalty b>-strangers, 55 percent to 71 percent of penalty The worst mistake, the execution would not reduce the number of erroneous the death sentences were returned 111 this or an innocent defendant, appears to be the convictions of that type, but rather -cvould comparati\7ely rare set of cases. rarest. This is ~vliatwe ought to expect: eliminate the worst consequei~cesol those The third step in the argument - that Guilty or innocent, few of those who are errors. The causes in the second group capital cases get an estraordinaiy amounr sentenced to death in America are actually imply that the deatli penalty i~self of attention - is also certainly true. But for executed. Among the knowin cases of undermines the accuracy of our system of the puipose of min~m~ingthe risk ol \vrongful conr.lction, many more innocent adjudication. As Justice Fi-ankf~lrcerput it: erroneous con~.ictionsand executions that defendants were either con~~ctedof first "'\Vlieii life is at hazard in a trial, it attention is a t\~~oedged sword at best: It degree murder and sentenced to cleat11 but sensa~ionalizesthe whole thing allnost generaies many more mistakes than we not executed, or convicted of first degree unwittingly The effect . . . [is] very bad." If ~17oulclsee if capital murders were handled murder and sentenced to life inipiisonment; that's true, abolislimg capital punishment as casually as run-of-the-mill robbelies and much smaller groups were convicted of would reduce the number of erroneous assaults. The extra attention we devote to second degree murder, or even man- convictions of all sorts in those cases in capi~alcases might also help us catch some slaugliter or lesser felonies, and sentenced \vl~icliwe now seek the death penalty, and or even most ol these mistakes, to the to terms of years. not merely limit the llann of those errors estent that we are comniitted to doing so. A conviction can be "~m-ong"in many that do occur. Uniortunatel;! recent hismi7 suggests that ways. It might be excessive - for example, our commitment to correcting deadly if the defendant is really guilty of second judicial errors is weak. degree inurder but was convicted of first I!. HOWoften are innocent people The last paragraph must seem very degree murder; or the jury miglit have sentenced to death? puzzling: Why \\~oulcladded altention been right to coiiclude that the defendant increase errors? And yet, that non-intuitive committed the fatal act, hut wrong to reject It2 ailybodyk guess how illany of the statement 1s the core of my argument. 1 will a defense of insanity or self-defense; or a 3,365 pilsoners on death row are innocent develop it later, aster defining my terins conviction that is factually accurate might oi the nlurders for which they were and offering a biief discussion of the large have been obtained in violation of tlie condemned. But we are beginning to be \,:oluine of evidence that has accuinulatecl clefenclantk coiistitutional rights. I'm not able to place a lower bound on how few it that mistake11 convictions In capital cases concerneel with any ol these tyipes of nzay be, and 11's quite a few The major do occur on a regular basis. Finally, J \\ill errors. I shall limit my locus to convictioizs worli in this area is a st~~dyof w-oiigf~~l review what we might do and what we in ol "the wrong person," a clelendant who convictions in "potentially capital cases" by fact do to in~niinizethese tragedies. did not do the act that caused tlze death or deaths for which he was convicted. Professors Hugo Bedau aid Michael Erroneous convictions (as I have clehned Radelet. The first p~~blisliedversioi~ of this I. Defining the issues. them) inay occur disproportioiiately olten woi-k appeared in 1987; it listed 350 such in capital cases for [wo types of reasons: wrongful coi7T'Ict10ns fl-om 1900 through 1985, including 139 death sentences ancl The archetypal capital case is a highly (1) because of factors that are com~?zonor 29 executio~~s.111 1992 Professors Bedau publicized prosecutio~zlor a hiiltal and inevitable in capital prosecutions, but that and Radelet, together wit11 Constance gcrl-y nlurcler, in yliic1-1the clefencla~ntis occur in other cases as well - for instance, Putnain, published their lindings 111 tlie tned, convic~ecl,sentenced to death, and tile lac1 that the crime involves homicide, book 111 Spite ~[Ilzizocence.By then the or that it was heavily publicized; or eventually esecu~ec!.Needless to say, most catalogue had beein extended to 416 capital cases differ II-om [his standard in (2) because of consequences that flow fi-om miscarriages of justice, from 1900 through errors that will eventually be discovered are homicides (of all sorts) make up a iraction 1990. Some of the cases on their list are not yet lznown. The average time to release of 1 percent of all arrests in this country, notolious and controversial, ivhcluding for the cases that Radelet and his colleagues and about 3 percent of arrests for crimes of several of the executions: Biuno list is 7.34 years; the median time is violence. Murder and 11011-negligent Hauptmann, Joe Hill, Nicola Sacco and between six and seven years. The death- homicide account for 1.3 percent of all Bartolomeo Vanzetti. For these cases, there row population in the United States has criminal convictions, about 7 percent of are other writers who maintain that the been growing steadily for decades; as a convictions for violent crimes, less than 3 defendant was in fact guilty But the result, many prisoners on death row have percent of all commitments to prison, and precision of Bedau and Radeletk judgment been there six years or less. (4) Some cases about 10 percent of commitments to in every case hardly matters; itk the overall in which innocent death row prisoners prison for crimes of violence. Death pattern that tells the story In the great have been released - perhaps most - are sentences account for about 2 percent ol all majority of their cases the error has been not in the sample. Over a quarter of the murder convictions, less than two-tenths of admitted or is beyond dispute. And even total number of cases (18168) are from 1 percent of all convictions for violent the disputed cases suggest that there are Florida; California, which has the largest crimes, and perhaps three hundredths of 1 severe doubts about the defendants' guilt death row in the country - 477 compared percent of all criminal convictions. In other - which in turn means that many of them to 389 in Florida - has only two cases; words, capital cases are heavily over- were innocent. On the other side, Bedau and Texas, which has executed more represented among known miscarriages of and Radelet excluded cases in which the prisoners than any other state - 144 justice - 5 to 1 or 10 to 1 or 100 to 1 or defendants may well have been innocent, compared to 39 for Florida - has only six. more, depending on which comparison if, in their judgment, the evidence of The reason for th~sdisproportion, as the seems most telling. innocence was not sufficiently convincing. authors point out, is that Professor Radelet Does this mean that miscarriages of In any event, a conlpilation such as this works in Florida and has maintained justice are more likely in capital cases than can only be a list of illustrations of the detailed data on every capital prosecution other prosecutions? I think so, for reasons I problem, not a catalogue of errors. As in the state. If there were comparable data will explain in the next section. But there is Bedau and Radelet readily admit, nobody for all death penalty states, or if there was a also an obvious competing explanation lor knows how many miscarriages of justice comprehensive registry of all death row this striking d~sproportioi~.Since we pay have gone entirely undetected. inmates released because of doubts about more attention to lzomicides than to other In 1996, Professors Radelet and Bedau guilt, the total of known cases would be crimes, and more to capital cases than to and William Lofquist published a third much higher. But these resources do not other homicides, we would be liliely to study on this issue: a compilation of cases exist. detect more errors among homicide of prisoners who were released from death The essential thing to know about convictions than among otlzer felonies - row since 1970 because of serious doubts mistaken convictions in capital cases is that and especially among the most aggravated about their guilt. They list 69 such cases, they do happen and will continue to homicides - even if the errors that occur about 1.2 percent of the total number of happen with some regularity - as Bedau were evenly distributed. In part, this death sentences returned between the end and Radelet halie shown. Bedau and argument is certainly true. With more effort of 1972 and the beginning of 1998. As the Radelet do not try to estimate how often we could discover more miscarriages of authors point out, their definition of the these tragic mistakes occur, and neither justice, and we do devote more attention to categoiy - "serious doubts about guilt" - will I. Instead, I nil1 address a related issue: capital cases than to other felony includes some death row inmates who Why do thej~happen in death penalty prosecutions. But it cannot be a complete were ultimately acquitted, or whose cases cases? explanation for tlhe apparent abundance of were dismissed, but who may in fact have At the outset, however, it may be useful errors in capital cases. Many of the known been guilty Nonetheless, it is almost to put the numbers I have provided in miscarriages of justice - capital and uon- certainly an undercount of the number of perspective. Bedau and Radelet have capital alike - were discovered by sheer defendants erroneously convicted and sent assembled information on more erroneous chance. If chance were the only factor, the to death row, for several reasons: (1) In conviclions in capital cases in Amelica in known cases would be representative oi all some of these cases - the most tragc - this century than all other collections of ei~oi-s;since it3 only one causal factor, the the error will never be discovered and the such errors in all criminal cases combined. sample is no doubt quite different from the defendant will be executed or die in prison Since then, similar errors keep coining to universe. Still, if even a third of the errors of other causes. (2) In others the error will light. In 1988, Arye Rattner published the suriaced by luck alone, it would be probably never be discovered because it most con~prehensivesummaiy of surprising if the actual proportion of errors has become moot. The published list does information on known miscarriages of in murder cases were over-represented in not include any case in which a defendant justice in Amenca, regardless of crime or the set of lznown errors by as large a factor who might well be innocent obtained cause - 205 erroneous conrictions, from as we see: live or ten or a hundred to one. release on other grounds, such as a 1900 on. In 45 percent of Rattnerk cases Ultimately, the comparative proportion constitutional violation, or the death or the offense was murder, and in 12 percent of miscarriages of justice in capital cases absence of a witness. (3) In some cases the penalty was death. B>7 compaiison, does not matter. ICs possible, I suppose.

Liwci QUADRANGLE NOTES SPRING1999 85 A. Investigation. This is the critical stage, where most errors occur. The circumstances that produce thern are variable, but the basic cause is the same: Homicides, and in particular capital homicides, are pursued much more vigorously than other crimes. As a result, more guilty defendants are identified and apprehended. Unfortunately, along the way more innocent defendants - a larger number and a higher proportion - are caught up in the process as well.

1. Clearance rates. Most crimes are never solved. In 1995, a mere 21 percent of all serious crimes known to the police were "cleared" - which usually means that a suspect was arrested; of serious violent crimes, 45 percent were cleared. But e\7enthese low figures only tell half the story. Most crimes are not "known to the police" - in 1995, only 36 percent of all crimes, and 42 percent of crimes of violence, were reported. In ot'her words, only about 18 percent of all crimes of vlolence are solved by the police, including about 14 percent of robberies, 18 percent of rapes, and 7 percent of burglaries. On the whole, the crimes that are reported to the police have better evidence than those that are not reported. Cases with extremely strong evidence - those in that erroneous convictions are just as Ill. Why are innocent people which the culprit is caught in the act, or common in other criminal cases. It's a regularly sentenced to death? seen and identified by several people - depressing thought. It implies that behind are almost always reported. If the victim has to take the initiative to notify the the seventy some prisoners who have been The road to conviction and sentence has police, he'll be more likely to do so if he released from death rowTin recent years three main stages: investigation, which is thinks there's a good chance that the because of doubts about their guilt there primarily the province of police; pre-trial criminal will be caught. When the police are thousands of undiscovered cases of screening and plea bargaining, where the do hear about a robbery, or a rape, or a defendants uith equally doubtful dominant actor is the prosecutor; and trial, burglary, for which the identity of the convictions for non-capital homicides, and before a judge and jury. At each stage, criminal is not immediately obvious, their dozens of thousands or more equally capital cases receive more care, more investigation is usually perfunctory: Put out questionable convictions for robbery, resources and more scrutiny than other a call to other officers to try to spot the burglar)! and assault. But even if we prosecutions. This special focus is a natural criminal in flight; interview the witnesses at assume this unlikely equivalence, the basic consequence of the unique importance of the scene; collect immediately available problems would be the same. Capital cases death - the deaths of the victims and the physical evidence; that's it. If a suspect are at least as error prone as any others prospect of death as punishment for the doesn't emerge from this process it is (if not much more so) and we regularly defendants. In most cases, the effects of unlikely that the case will ever be sentence innocent people to death. So the this special treatment are beneficial. But prosecuted. Most police detectives do not underl?ing question remains: Considering there's a cost: In some cases, the very same have the time to conduct detailed all the attention we devote to death penalty process produces terrible, deadly errors. investigations of every reported felony, and cases, why do we make so many mistakes? in the usual run-of-the-mill case there is conclusions, and - if they believe they error was a factor in only 16 percent of litile plessure on them to do so The net have the killer - perhaps to manufacture Bedau and Radelet's cases of errors in result is that in general the felonles lhal are evidence to clinch the case. The danger potentially capital prosecutions - which prosecuted are likely lo be those In which that the investigators will go too far is suggests that among the non-murder cases the emdence of gullt 1s strongesi magnified to the extent that the killing is in Rattner's sample, over 80 percent of tlhe Homicides are different First, almost brutal and horribng, and to the extent errors were due to misidentifications. every homicide is reporled to the pollce that it attracts public attention - factors No doubt the main reason for this when the body of the deceased person is which also increase the likelihood that the difference is the absence of a live victim in found Second, most homlcldes known LO murder will be treated as a capital case. most hoinicides. Victims provide crucial the police are cleared - 65 percent in The murder of l0-year-old Jeanine identification evidence in most robberies 1995, more in prevlous years Overall, the Nicarico is a good example. In February and rapes, and so they make most of the proportion of all homicides that are solved 1983 she was abducted from her home in mistakes, when mistakes are made. In the is about four times higher than the Naperville, Illinois, raped and lulled - a absence of a victim the police may have no comparable proportion for other molent crime of stunning brutality The murder eyewitness evidence, and therefore no cnmes A study of robbery investigations m was the subject of a long, frustrating, room for eyewitness error. Ths is hardly an Chicago in 1982-83, by Franklin Zimnng unsuccessful investigation - a humiliating advantage for accuracy Many perhaps and James Zuehl, promdes an excellent public failure. Thirteen months after the most eyewitness identifications of criminals illustration 13 percent of all robbenes murder - and less than two weeks before by strangers are accurate. Frequently they reported to the police were solved wthin the local prosecutor stood for reelection - are corroborated or lead to other evidence two months (including a somewhat lower three men were indicted: Rolando Cruz, that greatly reduces the likelihood of error proportion of robbenes wth lnjunes to the Alejandro Hernandez and Stephen Buckley - fingerprints, stolen property reliable vlctims), compared to 57 percent of Cruz and Hernandez were convicted and confessions, etc. In addition, for about half robbeiy killlngs This difference cannot be sentenced to death; their convictions were of all violent crimes eyewitness explained by supellor emdence - on the reversed by the Illinois Supreme Court. identifications are extremely reliable contrary, robbery homicides wll usually They were convicted again, but this time because the crimes were committed by have weaker emdence, since the mctim is only Cruz was sentenced to death. Again relatives, friends, or others who are known dead -but must be due to a systematic the convictions were reversed. Finally at to the victims. Murderers are even more difference in the investigation by the police CruzS third ttial- over 12 years after the likely to be known to their victims but that As we have noted, many homicldes are murder - the case fell apart when a police may not help because, in the words of the easy to investigate In a typical case - a officer admitted he had lied under oath, immortal cliche, "dead men don't talk." kllling by a fnend as a result of a drunken and the judge entered a judgment of Eyewitness identifications are also very fighr - the killer is known from the start acquittal. What seems to have happened is unco~nmonin burglary cases, but the But the police get the hard murders as well this: Under intense pressure, the police upshot is different. There are very few as the easy ones, and theie is much more convinced themselves that they knew who erroneous convictions based on piessure to solve these cases than non- lulled Jeanine Nicarico and they misidentifications, but since there are also homicidal cnmes The relatives of the manufactured evidence to convince few burglary prosecutions based on non- vlctim care more, the prosecutol cares prosecutors and to use in court. If the eyewitness evidence, there are few ei-rors of mole, the public is much more likely to be criminal had taken jewelry from the any sort, and few convictions. The concerned, and the pollce themselves care Nicarico home rather than a child - or clearance rate for reported burglaries is more Death produces strong reactions - even if he had knocked out a family only 13 percent. But killers must be m his context, a desire to punish and to member or set the home on fire - there pursued, and in the absence of eyewitness protect Other outrageous crimes can have would probably have been a minimal evidence, the police are forced to rely on the same effect - kldnappings, for investigation, no arrests, no trial, and no evidence from other sources: accomplices; example, or senal rapcs - but they are erroneous convictions. jail-house snitches and other undenvorld lare Hoinicide 1s common figures; and confessions from the Foi the most part, the pressule to solve 2. Evidence. defendants themselves. Not surprisingly, homicldes produces the intended iesults Most miscarriages of justice are caused perjuly by a prosecution witness is the An lnvestlgation that would be closed by eyewitness midentifications. In Rattner's leading cause of error in erroneous capiral wthout arrest if ~t were a mere robbery sample of wrongful convictions, 52 percent convictions, and false confessions are the may end in a conviction if the robber killed of the errors for which the cause could be third most common cause. one of his vlctims But that saine piessure determined were caused by misidenti- Peijuty. From Macbeth to Mark Twain's can also produce mistakes II [he murder fications, and other researchers concur that Injun Joe, the killer who blames his crime cannot be rcadlly solved, the pohce may be eyewitness eri-ol; is by far the most on others is a familiar character in liction. tempted to cut corners, to jump to common cause of convictions of innocent Similar things happen in life. Some defendants. On the other hand, eyewitness criminals iinplicate innocent defendants in Carolina based in part on testimonj~from was convicted and sentenced to death order to divert suspicion from themselves. his cellmate, who said that Langley had almost entirely on Dunbar's testimony He In other cases, false nrltnesses, who may confessed to him. Langley came within half steadfastly maintained his innocence, and have had no role in the crime, lie for an hour of electrocution, but was claimed that Dunbar himself - a man money or for other favors. Both types of esonerated four years later and received a with a long history of violence and motives are more powerful in homicides full pardon. His cellmate didn't have to deception - must have committed the than in other climinal cases, and especially wait that long; after his pejurious murder. After the trial, Dunbar, who was in capital homicides. testimony, unrelated charges against him soon jailed for other crimes, bragged that First, the threat of being caught is much were dropped. he had framed Carrigei-. In 1987 he greater for a homicide than for almost anjJ Fifth, it's easier to lie about a capital case confessed his own guilt to various people, other crime. It's no news that the police than most other cnmes of violence: therek including his parents and a clergyman. work much harder to find killers than usually no live vlctim to contradic~the false That same year he repeated his confession burglars or robbers, and that their interest witness. in court, and admitted that he had lied at increases in proportion to the brutality and The overall result seems to be that Camger's trial and that he had committed notoriety of the crime. witness perjury is a far more common the murder himself. Three weeks later he Second, if the culprit is suspected and cause of error in murders and other capital retracted that confession, but admitted that caught, he has more to fear in a capital cases than in lesser crimes. Bedau and he was doing so for fear that he'd be case. He might get esecuted. The threat of Radelet identified it as a factor in 35 percent prosecuted for the murder and executed death can be a powerful motivator when of their erroneous capital convictions, himself. In 1991, shortly before he died in it's concrete. The death penalty as an while Rattner lists perjury as the cause of prison, Dunbar confessed again, to his abstract prospect does not seem to deter only 11 percent of his errors. But recall that cellmate. Dunbar's es-wife, who had many homicides. Before the crime, the 45 percent of Rattner's cases are murders. If corroborated his original story and had killer - if he thinks at all - no doubt pe jury were as common among the given him an alibi, testified in 1987 that expects to escape scot-free; he is not likely murder convictions in Rattner's sample as Dunbar had forced her to lie. to weigh the benefits of murder against the among Bedau and Radelet's cases, then In December 1997, the Ninth Circuit costs of the possible punishment. After the erroneous murder convictions could easily Court of Appeals en banc ordered that crime, however, there is more time to account for all the cases in which the error Carriger be retried or released. As of this think, and the fear of conviction and was caused by perjury. writing, he remains in custody awaiting execution may be vivid - especially if the The case of Paris Calriger is a good retrial. He came close to execution on police seem to be closlizg in. illustration of the role of pe rjuiy in capital several occasions in the 20 years since his Third, a pe junous killer may have to prosecutions. On March 14, 1978, Camger arrest. Under the circumstances, a new trial admit to crimes himself. He and the was arrested for the bi-utal robbery murder seems a modest goal, since, at a minimum, innocent defendant may in fact have been of Robert Shabv, the owner of a jewelry the evidence that has turned up after tilal accomplices in some crime other than the store, on the previous day The evidence raises grave doubts about Carriger's guilt. murder, or he might have been caught in against Cariiger was provided by Robert But if Robert Shaw hadn't been killed, none undeniably compromising circumstances, Dunbar, a friend on whose property of this would ever have happened. Dunbar or he might have to admit to some level of Camger was living in a trailer. Dunbar - would probably never have approached the guilt in order to make his accusation who had a great deal of experience as a police, they would hardly have given an credible. If so, the real killer has more to police informant - called the police and ex-felon immunity from prosecution for gain in a capital case than under other said he could identify Shaw's killer in three serious felonies in order to convict circuinstances. If he has to go to prison, return for immunity froin prosecution lor someone else of a single robbery, and the the gain from cooperation is tiine vs. death, various felonies: another robbery he victim would have been available to as opposed to less time vs. more tiine. But committed two days earlier, possession of a contradict a false story that may not be necessary: If he helps gun he had bought (which was illegal False confrssioi~s.A typical robbery break a capital case, he may walk. because he was a convicted felon), and investigation is resolved by an eyewitness Fourth, if the witness is lying to get attempting to dispose of the proceeds of identifica~ion;a typical homicide favors unrelated to the criine at issue, he'll the Shaw robbery-murder. The police investigation is resolved by a confession. do much better if it's a big case - whlch agreed to ihese terms. Dunbar then told Many confessions are easy straight-forward usually means a murder, or better yet, a thein that Carriger had come to him, affairs - volunteered by suspects who are capital murder. The typical witness in this confessed to the lulling, and asked for help overcome by guilt, or believe they have categoiy is the jail-house snitch. For in disposing of bloody clothes and stolen nothing to lose. These are he easy cases, example. in 1932 Gus Colin Langley was jewelry; Dunbar corroborated the story by where nothing has been clone that i~zight con~ictedof fil-st degree murder in North producing some ol the loot, and leading produce a false coi-~fessioiz,and where the police to some of the cloihes. Camger mole often ~hannot theie is stlong The case of Melvin Reynolds is a good police investigation, but il has more impact coiloborating evldence of gull[ Some example, but by no means unique. On on the accuracy of ciiminal dispositions coiifessions, however, are not so readily May 26, 1978, 4-year-old Eric Christgen than anything that happens later on. If (he given, but ale instead the end products of disappeared in downtown St. Joseph, wrong person has been arrested, this is long, drawn out interrogations Missouii. His body later turned up along where the mistake is most likely to be Ameiican police oificels use all soits of the Missouri fiver; he had been sexually caught. But in capital cases the value or coercive and manipulative methods to abused and died or suffocation. The police that screening is undermined, in part by obtain coiifesslons They confuse and questioned over a hundred possible the effect of the threat of the death penalty disonent the suspect, they lie to hlm about suspects, including "every known pervert and in part by the attention and pressure physical evldence, about wtnesses, about in town," to no avail. One of them was that capital cases generate. As a result, there statements by other suspects, they pretend Melvin Reynolds, a 25-year-old man of is a danger of two distinct types of errors. that they already have thelr case sealed and limited intelligence who had been sexually ale only givlng the suspect a chance to abused himself as a child and who had 1. Guilty pleas by innocent defendants. explain his side of the story, they pretend some homosexual episodes as an Threat is an essential part of all plea to understand, to sympathize, to excuse, adolescent. Reynolds, although extremely bargaining: Take the deal or you'll do they play on the suspect's fears, hls biases, agitated by the investigation, cooperated worse after conviction. There is, his guilt, his loyalty to family and fnends, through several interrogations over a undeniably, a coercive aspect to thls his rehglon, they exhaust the suspect and period of months, including two polygraph bargain - the defendant must risk a severe wear him down, 111 some cases, they use examinations and one interrogation under penalty in order to exercise his right to trial molence, even toltule These are powerful hypnosis. In December 1978 he was - and plea bargaining has been strongly techniques They work to get confessions questioned under sodium amytal ("truth criticized on that ground. One attack is from gullty defendants - and sometimes serum") and made an ambiguous remark that the threat is so effective that it drives from innocent defendants as well that intensified police suspicion. T~vo sonie innocent defendants to plead guilty From the pomt of mew of the police, months later, in February 1979, the police along with the mass of guilty ones. That the inain problem w~hinLerrogatioil is not brought the still cooperative Reynolds in may happen with some regularity for that it occasionally pioduces errors, but for another round of interrogation - 14 innocent defendants who are offered very that ~tkextremely time consuming It's hours of questions, promises and threats. light deals: time-served, diversion, six likely to take hours, peihaps days to break Finally, Reynolds gave in and said, "1'11 say months unsupervised probation, and so down a suspect who resists and inslsts on so if you want me to." In the weeks that forth. But among the more serious criminal 111s innocence Frequently several police followed, Reynolds embellished this con~ictionswith severe penalties of ofhccrs cooperate in the effort, ques~ioiiing confession with details that were fed to imprisonment or death - those the suspect simultaneously or in relays As him, deliberately or othenvise. That was convictions that show up in cases of a result, extended interrogation is largely enough to convince the prosecutor to proven miscarriages of justice - the iesenred for big cases in whlch confessions charge Reynolds, and to convince a jury to picture is different. I have located exactly are necessaiy for successful prosecution convict him of second degree murder. He one reported miscamage of justice based Tjrpically, [hat means homlcidcs, and was sentenced to life imprisonmenL. Four on a gullty plea for a non-homicidal cri~ne especially the nios~heinous homicides, for years later, Reynolds was released when - and tliar was a peculiar case, a reasons Itre mentioned thcse are the cases another man - Charles Hatcher - defendant who pled guilty to a criine he that the pollce are most annous to solve, confessed to three murders, including tliat did not commit along with one which he and yet, because the vlctim is dead, they of Eiic Christgen. did commit. The available collections of frequently lack cycmiltnesses known errors are hardly representative As bmth pequry, false confessioiis ale a 0. Pre-Trial Screening samples oi the universe of erroneous much more common cause of elrors foi Most prosecutions are resolved without convictions, and ei-rors based on guilty homicides than for other cnmes They tiial. Eighty to 90 percent of convictions pleas are undoubtedly less likely to be weie a cause of 14 pelcent of Bedau and result from guilty pleas, usually after plea discovered than those based on trials. Even Radelet's errors in homiclde and capltal bargains, and at least 80 percent of so, this is a stark contrast to the cases, but only 8 perccnt of the errors defendants who are not convicted obtain ovenvhelmiiig proportion of all conrictions reported by Rat~nerSlnce 45 pelcent of pre-tiial dismissals rather than acquittals. that are based on guilty pleas. Rattnel's cases arc homicides, this suggests In otliei- words, most of the work of sorting Judging fro~nthe available e~ridence, that false confessions are three to four criminal cases after arrest is done pre-trial, innocent defenclanls rarely plead guilty times more common as a cause of by the exercise of prosecutorial discretion when doing so entails a substantial tet~nof mlscaiiiages of justice for homicide cases to dismiss, to reduce charges, or to imprisoumeni, except in capital than foi oiher cnmcs recommend or agree to a particular prosecutions. Radelet, Bedau and Putna~n sentence. This pre-trial screening is list 16 cases of innocent lzo~nicide undoubtedly less important than the initial defendants who pled guilty; in most, fear may be dismissed even if the prosecutor is It is bound to be much more difficult - of execution is given explicitly as the convinced oi the defendant's guilt. Regard- and unlikely - if the crime has attracted a reason for the plea. This is, no doubt, less of their belief in the defendants' guilt, lot of attention, or if a victim, or several, another illustration of hoar deach is prosecutors focus on the easiest cases first were killed. different. It seems that innocent defendants - the ones with the best evidence - since The problem is not just public pressure. will almost always risk additional years of those are the cases where their limited Evidence of a defendant's innocence does their lives in order to seek vindication resources will have the greatest impact. not arrive on the prosecutor's door step on I-ather than accept disgrace coupled with a But homicides are different. Homicides its own. If the police didn't find it at an long tern1 of imprisonment, but some will (and other notorious crimes) are the cases earlier stage, it is usually presented by the not go so far as to risk death. for which resources are consewed. A dead defendant's attorneys. Everybody agrees The case of John Sosnovske is a good loser will still be dismissed, but what if it's that innocent defendants should not be example. In 1990, he was falsely merely likely that the defendant will be charged or convicled; the trouble is implicated in the rape murder of Tauizja acquitted? If it's a robbery, the prosecutor identifying the cases in which that applies. Bennett by his grl friend, Laverne Pavlinac, may dump the case and try another; if it's a If there happens to be overwhelming who apparently was afraid of hiin and murder, she's more likely to forge ahead. independent evidence of innocence, there anxious to be lid of him. In the process, Prosecutors lose a much higher is no problem. But if the evidence of the Pavlinac became entangled in her own lies, proportion of murder trials than other defendant's innocence is not so clear, or if and claimed to have participated in the felony trials, about 30 percent vs. about 15 its significance is not obvious, the killing. Both were charged with murder. percent. As Robert Scott and William defendant's fate may hinge on the Pavlinac recanted her confession but was Stuntz point out, the most likely prosecutor's willingness to listen with an convicted and sentenced to life in prison. explanation is that in murder cases they are open mind. The more notorious the case, Following her conviction, Sosnovske - willing to go to trial with comparatively the more difficult that may be. Prosecutors, who was facing the death penalty - pled weak evidence. The main effect of this like the rest of us, have a harder time no contest and was also sentenced to life extra effort is that guilty defendants are recognizing an error the more publicly they imprisonment. Both were freed in 1995 convicted who otherwise would never be have endorsed it, and the more time and alter another man, Keith Hunter Jesperson, tried. But in some cases the evidence is money and prestige they have committed confessed and also pled guilty to the same weak because the defendants are not guilty, to it. murder. and some of those innocent defendants are A prosecutor can always discount the not only tned but convicted. In other defense attorney's claim that her client is 2. Failure to dismiss false charges. words (as with police investigations), as innocent: This is hardly a non-partisan The major filter that may prevent a prosecutors work to obtain convictions in source. An attorney for an innocent charge based on questionable evidence hard homicide cases they draw in cases defendant must overcome this handicap in from turning into a conviction is where it's difficult to separate the innocent any case; in capital cases it may be pi-osecutorial discretion to dismiss. Overall, from the guilty insurmountable. In an ordinary criminal dismissals of felony charges outnumber Prosecutors also disiniss charges in case, most pretrial contact between the acquittals about 4 to 1. Many cases are some cases because they believe the prosecutor and the defense attorney takes dismissed because of weak evidence defendant may be innocent, regardless of place in the context of plea bargaining. But despite the fact that the prosecutor is the evidence that is available to obtain a in many capital cases - especially those convinced that the defendant is guilty; conviction. The rules of professional most likely to produce death sentences - other cases are dismissed because the responsibility allow a prosecutor to there is no plea bargaining. The prosecutor prosecutor is convinced of the defendant's consider her own view of the defendant's knows from the start that she will insist on innocence, or has at least come to doubt guilt in deciding whether to charge, but do the death penalty, so there is nothing to his guilt. For homicides, and especially not require her to do so. Prosecutors have bargain over. In the absence of plea capi~alhomicides, both sorts of dismissals widely varying views on how to apply this bargaining, there will be fewer open are less likely. In both situations, the major vague standard, from those who say that channels of communication between the reason is the same: We devote more they will never prosecute unless they defense and the prosecution, so it may be attention and more resources to criininal themselves are convinced beyond a harder for the defense attorney to get a cases when death is at stake. reasonable doubt of the defendant's guilt, serious hearing. Worse, in that context, the Trials are time consuming and to those who believe that regardless of their true value of a claim of innocence becomes expensive; they are a scarce resource. Since own uncertainty their task is to make a harder to interpret. When plea bargaining most cases cannot be tried, it is obviously case and let the jury decide. But this is is an option, a defense lawyer is not likely sensible for a prosecutor to tq7 to restrict always a discretionary choice, and to commit her credibility to the argument t~ialsto cases where the outcomes will be whatever the prosecutor's position in the "He didn't do it" unless the lawyer believes useful - i.e., convictions. If possible, a abstract, an actual decision to disiniss a that it's true, since (quite apart from likely loss at trial nil1 be avoided through serious charge that would probably have possible effects on her reputation) taking generous plea bargaining; if not, the case resulted in a conviction is always difficult. that position will undermine her ability to do juries spot innocent defendants that the percent that went to jury trial. In other comparatively minor role in the production prosecutors have missed? Unfortunately, words, since pre-trial sorting does less to of errors in capital cases. To the extent that juries approach this task with two severe winnow homicide cases than other jury behallor at trial does matter, the handicaps: They have less information than prosecutions, homicide defendants are more question is: Do juries behave differently in the prosecutors or the police, and they likely to face the chancy ordeal of trial. homicide trials in general, and in capital have essentially no e'xperience. Given these 1 don't mean to say that the institution homicides in particular, than in other limitations, it is unrealistic to esFect juries of trial by jury does not help reduce the criminal trials! There are several reasons to to systematically correct errors in the incidence of erroneous convictions. It no think that juries treat homicides and capital earlier decisions to investigate, to arrest and doubt does fill that function, but by brute cases differently than other criminal cases, to prosecute. force: by making it more difficult for the and most of them point in the direction of This is bad news for homicide prosecution to obtain any convictions, and a higher likelihood of conviction. defendants. Whether it's because by discouraging trials of the guilty and the prosecutors take weaker cases to tiial or innocent alike unless the evidence of guilt 1. Factors that increase the likelihood because they insist on the maximum is very strong. The main benefit of this of conviction. I penalty, homicide defendants are more process is that feedback from court may Publicity. Most crimes, even most likely to face juries than other criminal improve pre-trial investigations and homicides, receive veiy little attention from defendants. For example, in 1988, 33 increase selectivity in charging - the the media. A few crimes, however, are percent of murder cases in the 75 largest stages of the process we have already heavily publicized. Many, perhaps most of counties in the United States went to trial, discussed. If all works well, the result is these notorious crimes are homicides, and compared to 5 percent of all felony that few innocent defendants are brought especially the unusual and heinous prosecutions and 9 percent of all violent to trial, most defendants who are convicted homicides that are most likely to be felonies. In 1994, 15 percent of robbery are guilty and most who are acquitted are charged as capital crimes. In those cases, convictions across the country were also guilty. And yet, if an innocent most jurors will have heard all sorts of obtained at trials, of which 10 percent were defendant is tried, he will probably be things about the case before they got to jury trials, while 42 percent of murder convicted. court, many of them inadmissible, convictions were after trial, including 35 Given this structure, trial plays a misleading, and inflammatory They may have seen or heard or read that police and force it to spread its resources more 38 [1980]),the United States Supreme officers 01- other government officials have thinly This distraction might increase the Court acknowledged this possibility and declared the defendant g~~iltyThey may chances of consiction even lor those capital held that a juror could not automatically be have witnessed or felt a general sense of defendants who are represented by skillful excluded from sellice because of this communal outrage. All this will malze them la~i3rerswith adequate resources; it will be reaction. To the estenl that jurors do feel more likely to convict. Courts may attempt far more damaging for the many capital this way, they may be less likely to convict to mitigate the impact of pre-trial publicity defenclants whose defense is shamefully in capital trials than in other hon~icides. by various means - most effectively by inadequate. changing the location of the trial - or they Heilzous~zess.In theory, jurors are 3. Net effects. may refuse to clo so. Not surprisingly, the supposed to separate their decision on the When there are forces that push in one records of erroneous convictions include defendant's guilt from their reaction to the direction and forces that push in the other, scores of cases in which publicity and heinousness of his conduct: If the evidence it is sometimes possible to say that they public outrage clearly contributed to the is insufficient, they should be just as cancel out. Not here. The effects I lhave error - from the convictions of Leo Frank willing to acquit a serial murderer as a described are extremely variable. Publicity, in 19 13 and the Scottsboro Boys in 1931, shoplifter. Nobody believes this. Even in death qualification, the lheinousness oi a to the con~~ctionsof Rolando Cruz and civil trials, where the jury is asked to homicide - each of these may make a Alejandro Hernandez in 1985. decide cases by a preponderance of the critical difference in a particular case, or it Death Qualgicatiotz. In capital cases, evidence, there are indications that juries may not. On the other side, the protective juries decide the sentence as well as (and judges) are more likely to find features of capital trials are uneven at best. determine guilt or innocence. To defendants liable, on identical evidence, as Many capital defendants do not have accommodate this function, the capital jury the harm to the plaintiff increases. In quality representation, by any standard. selection process includes a unique criminal trials the problem is worse, since And the anxiety that jurors may feel when procedure, "death qualification," that is the burden of persuasion is proof beyond a a defendant's life is at stake will be relleved designed to ensure that the jury is qualified reasonable doubt. In a close criminal case if a jury decides (as they may do in for the sentencing phase, hlost jurors a7ho the jury is supposed to release a defendant deliberations on guilt) that he will not be are strongly opposed to the death penalty, who is, in their opinion, probably guilty sentenced to death. With that out of the and some who are strongly in favor, are This is a distasteful task under any way, the competing rmpulse - to not free excluded at the outset. Many studies have circumstances, but it becomes increasingly a man who has killed - may take over, in sho~mlthat these exclusions produce juries unpalatable - and unlikely - as we move force. that are more likely to convict. In addition, up the scale from non-violent crime, to I once saw a cartoon of two men in the process of questioning jurors about violent crime, to homicide, to aggravated black robes, obviously judges, talking in a their willingness to impose the death grisly murder. hall. One says, "Some days I'm feeling good penalty before the trial on gurlt or and everyone gets probation, and some innocence has begun, tends to create the 2. Factors that decrease the likelihood days I get up on the wrong side of bed and impression that guilt is a foregone of conviction. I throw the book at eveiybody It all conclusion, and the only real issue is Quality of Dejense. Capital defendants, balances out." In statistical terms, the punishment. and to some extent homicide defendants in problem is increased valiance: Since Fear oJ Death. In a capital case, avoiding general, may be better represented than nobody gets the average punishment, the execution can become the overriding other criminal defendants. The attorneys more the judge's sentences are spread out imperative for the defense. In extreme who are appointed to represent them may arbitrarily, the more of them are errors - cases, fear of death drives innocent be more experienced and skillful, and their and errors on one side don't balance out defendanls to plead guilty in return for a defenders may have more resources at their errors on the other. The same is true of lesser sentence, even life imprisonment. If disposal. Other things being equal, higher decisions on guilt and innocence: Mistakes the defendant does not plead guilty, either quality representation will decrease the in one direction in some cases do not because no plea bargain is offered or likelihood of conviction, and may operate balance mistakes in the opposite direction because he was un~villingto take it, the as a check on errors and misconduct that in other cases. In capital trials, one same pressure ivill be felt at trial. Fear of a drive some innocent capital defenclants to particular type of ~nistalte- conviction of death sentence may drive he defense LO trial and to conviction. an innocent defendant - is ovenvhelm- lnake tactical choices that compromise its Severity oJ the Pa~alty.Prosecutors, ingly important, and the fac~that other, position on guilt in order to improve the defense attorneys, and judges widely guilty defendants get [he benefit of other odds on penalty; in some cases, he defense believe that some jurors are more reluctant errors is no help. If you're building a may virtually concede guilt and focus LO convict a defendant who might be seawall, adding height to one part won't entirely on punishment; iL will certainly execuied than one who faces a less extreme make up for cutting away at another. distract [he defense from the issue of guilL punishment. In Ada~qzsv. Texm (448 U.S. IV. Conclusion: catching errors.

The basic conclusion is simple. The steady stream of errors that we see in cases in which defendants are sentenced to death is a predictable consequence of our system of investigating and prosecuting capital murder. And behind those cases, there is no doubt a larger group of erroneous convictions in capital cases in which defendants are not sentenced to death. But what about what happens after trial? Everybody knows that direct and collateral review are more painstaking for capital cases than for any others. Isn't it likely that all these mistakes are caught and corrected somewhere in that exacting process? The answer, I'm afraid is, No. At best, we could do an imperfect job of catching errors after they occur, and in many cases we don't really try As a result, most miscarriages of justice in capital cases never come to light. Probably the best way to figure out how to catch miscarriages of justice is to look at the cases in which we have done so. Judgng from the cases that are reported, three factors, separately or in combination. are usually responsible for an innocent defendant's exoneration: Attention, Confession, and Luck. Attention. If a defendant is sentenced to death, he may well get more careful and attentive consideration from the courts on rev-ie~v.hMore important, he is likely to be better represented on direct appeal than he would be othenvise, and he is likely to have counsel on the post-appellate collateral review, while most defendants have none. These advantages may explain in part the high proportion of death sentences among known miscarriages of justice. But a comparative advantage is not a panacea. Many death row inmates have inadequate representation at e17et-ylevel of review, and some have no legal assistance whatever for collateral review. And many capital defendants who are convicted in error are not sentenced to death, yen likely most. They do not recei1.e any special attention from their attorneys or from the courts; on the contraq they might suffer from the perception that they've already received the benefit of whatever doubts their cases may raise. When \Valter what to do: stop the execution, release the prisoner. If there were some general method for identifying mistakes, we ~~uldn'thave this problem in the first place. But of course, there isn't. Instead, the errors that we have discovered advertise the existence of others that we've missed.

McMillian was released after six years on It works: More cases are cleared, more throws the entire weight of detecting errors death row for a murder for which he had murderers are convicted. But harder cases onto the re\iewmg courts: since the been framed by local enforcement officials, are more likely to produce mistakes - still discovery of errors takes time, the main his attorney said that "only the death exceptions. no doubt. but not as rare as for burden is on the later stages of the process, sentence had allowed Mr. McMillian to other crimes, where the cases that are and especiaily hahcas corp~isreview in the receive adequate representation, which prosecuted are mostly skimmed off the top. federal court. Recently, resources for post eventually uncovered the plot against him." Perhaps the worst mistake we might make conviction defense in capital cases have In truth, McMillian's post-conviction in this connection is to assume that the been cut, the bases for review in federal representation was not adequate, it was danger of error for homicides is as small as court have been limited, and the process of extraordinary If he had merely been it is for other crimes, or, worse yet, that it review has been accelerated. If a defendant sentenced to life imprisonment, he may is even smaller. Homicides, especially obtains evidence of his innocence late in

mention relief, are extraordinarily high. Confessions. In most cases in which Perhaps these new rules will have little miscarriages of justice are uncovered, the discovered and the mistake is proven effect in practice. But if they do, the real criminal confesses to the crime. In the beyond doubt, we know what to do: stop direction of change is inevitable: Fewer common scenario, the true murderer is the execution, release the prisoner. If there mistakes will be caught even among those arrested and imprisoned for another crime were some general method for identifyng cases that remain on track to execution, - sometimes a similar homicide - and mistakes, we wouldn't have this problem in more innocent homicide defendants will confesses before trial or in prison. For the first place. But of course, there isn't. remain in prison, and more defendants will example, Mehin Reynolds confessed Instead, the errors that we have discovered be killed by the state in error. falsely under intense pressure, to the rape- advertise the existence of others that we've murder of a 4-year-old boy; he was missed. How often will an innocent released when Charles Hatcher was prisoner run into a movie producer who is arrested and confessed to three murders, struck by his story? What if the real killer R-Gross, the Thomas and Mabel including the one for wh~hReynolds was is killed in a car crash, or dies of a drug LOng Professor of Law* is a recognized imprisoned. Similarly,John Sosnovske and overdose, or is never arrested, or never authority on the death penalty and has written Laverne Pavlinac were both freed in 1995 confesses?The most the legal system can widely on the snbject. He also has published on after Keith Jesperson confessed to the murder do is improve the odds by providing eyewitness identification, the we of expert for which they were falsely convicted. resources to help discover and prove witnesses, and the relationship between pre- Luck. Getting a confession from the real errors, by considering serious claims trial bargaining and trial verdicts. A graduate killer is the common stroke of luck in cases whenever they are made, and by talung of Columbia college^ he earned hisj.D. at the in which a miscamage of justice is caught. action even if proof of innocence is not University of California at Berkeley. He was in But sometimes luck takes a different route. absolute. private practice in San Francisco and ~vorked The break in Randall Dale Adarns' case Attention and quality representation as an attorney with the United Famz Workers came when documentary film maker Errol improve an innocent defendant's chances. Union?the Knee Lega2 Defense' Moms ran into Adam5 by chance in 1985 They help get court hearings: they increase Offense the NAACP Legal Defense and Educational Fund, lnc., and the National when Moms was doing research on visibility, which produces opportunities for psychiatric testimony in Texas capital lucky breaks; they buy time during which ILLYPr@kct before going into teaching. He prosecutions. Moms went on to produce a the true killer may confess. But these teaches in the fields of evidence, crimina2 procedure, and the usc of social sciences in law movie about Adams' case, The Thin Blue assets, whatever their value, are unevenly Linc, which was released in 1988; the distributed. For the most part, they are the mo~iedrew national attention to the case special preserve of defendants who have and resulted in Adams' release in 1989, 12 been sentenced to death and who still face years after he had been sentenced to death. the possibility of execution. And even for The basic cause for the comparatively that restricted group this special attention large number of errors in capital cases is a is under fire. Executive clemency - the natural and laudable human impulse: We traditional backstop that was said to revent execution "when there is the

94 THE UNIVERSITYOF MICHIGAN LAW SCHOOL