Main Content

Advanced Evidence Spring 2023 (4 credit)

Blood Will Tell (Part 2)

This is part two of a two-part investigative journalism report about the trial of Joe Bryan. Joe was connvicted of murdering his wife Mikey on the basis of blood spatter evidence. The link leads you to full article (which includes photos), but you only need to read the following sections, which focus on the bloodstain pattern evidence:

II.

LAST FALL, more than three decades after Joe’s conviction, I found myself surrounded by human blood. I had signed up for a class in bloodstain-pattern analysis in the hope of gaining a better grasp of both bloodstain interpretation and the training police officers receive in the discipline. At each of Joe’s trials, prosecutors used the testimony of Robert Thorman, a bloodstain-pattern analyst, to lend scientific authority to an ambiguous case, and I wanted to more fully understand the basis of his expertise.

Thorman’s testimony had been critical, because the state’s theory of the case posited an extraordinary sequence of events. Prosecutors asked the jury to believe that between 9:15 p.m. on Oct. 14, 1985, when the Bryans spoke by phone, and the following morning, when Mickey was found shot to death, Joe slipped out of his hotel in Austin; drove 120 miles to Clifton, at night, through heavy rain, even though he had an eye condition that made night driving difficult; shot his wife, with whom he had no history of conflict; drove 120 miles back to Austin; re-entered the hotel; and stole upstairs to his room — all in time to clean up and attend the conference’s morning session, and all without leaving behind a single eyewitness.

The main piece of evidence they had was the blood-speckled flashlight that Charlie Blue found four days after the murder in the trunk of Joe’s car. What connection it had to the crime, if any, was unclear; the blood on it was type O, which corresponded not only to Mickey but also to nearly half the population. To secure a guilty verdict, the prosecution needed to tie the flashlight to the crime scene. With the unassailable certainty of an expert, Thorman testified that the flecks of blood on the flashlight lens were “back spatter” — a pattern that indicated a close-range shooting. He wove a narrative that placed the flashlight in the killer’s hand. He also allowed the prosecution to explain away the lack of blood in the interior of Joe’s car when he asserted that the killer had changed his clothes and shoes before fleeing the house. Thorman’s testimony was decisive, making the state’s tenuous theory seem plausible. But jurors at the first trial did not know that, at the time, the only formal training Thorman had in the forensic discipline was a weeklong class. He took it four months before Mickey’s murder.

Thorman took his class in Beaumont, Tex., in 1985, but such classes are still offered at Police Departments across the country — including Yukon, Okla., where I was taking my bloodstain course. Around me, drying on long sheets of butcher paper that hung from the walls and unspooled across the floor of the Police Department’s garage, were the common types of bloodstains that might be found at a crime scene. Under the direction of our instructor, my classmates and I created each pattern in a different way: by swinging a bloodied ax through the air, or pressing a blood-soaked knife to a piece of cloth, or firing a pellet gun at a bloodied sponge. Lumbering around the garage in the biohazard gear we were provided — hooded white Tyvek coveralls, latex gloves, safety goggles and masks secured with duct tape — we tried to classify each bloodstain according to a dizzying taxonomy of spatters, drips, spurts, swipes and smears.

The course, which cost $655, was offered by Bevel, Gardner & Associates — the consulting firm of Tom Bevel, one of the discipline’s most sought-after expert witnesses and a co-author of its principal textbook, “Bloodstain Pattern Analysis With an Introduction to Crime Scene Reconstruction.” Bevel served as Thorman’s instructor back in 1985, and it was from Bevel that Thorman acquired his understanding of the discipline. Though Bevel no longer teaches, his reputation’s reach was reflected in the makeup of my class. There were Oklahoma police officers, Oregon forensic scientists and two law-enforcement agents from Taiwan. Our instructor was Tom Griffin, a partner of Bevel’s who spent 27 years with the Colorado Bureau of Investigation.

The first two days of our class were largely devoted to identifying the different types of bloodstain patterns. We learned that a line of circular or elliptical stains leading across a floor was a “drip trail”: a pattern created when, say, a wounded victim staggered across a room. Many patterns were far less distinct, though, and as we scrutinized them, I came to realize an unsettling fact: Two bloodstains can look nearly indistinguishable from each other, even if the actions that created them were wildly different. Blood, when expelled from the mouth or nose, for instance, may appear almost identical to the atomized blood that sprays from a wound when someone is shot. In real life, a crime scene would provide context that would guide our conclusions, but as we studied the nuances of different bloodstain patterns, I was struck by how open to error this enterprise was.

As the week progressed, we were instructed to do increasingly complex work with little understanding of the trigonometry and fluid dynamics involved. We were taught how to calculate the area of origin: the place, at a crime scene, where the bloodletting originated. (In a shooting, this would be the three-dimensional area in space where the bullet pierced the victim’s body.) Armed with calipers, scientific calculators and string, we measured bloodstains, plugged our data into equations and tried to trace the trajectories of individual droplets back toward their source. As was true with pattern classification, there were many ways to get this wrong; small deviations with the calipers resulted in markedly different results. Still, Griffin had us press forward. “We’re not really going to focus on the math and physics; it just kind of bogs things down,” he told us at the outset. “I’ll teach you which keys on your calculator to press.”

It was upon this shaky foundation that Thorman had tried to reverse-engineer the shooting at the Bryan home. Looking over his 1985 report one night back at my hotel, I could see where his analysis went awry. According to his report, he believed he was determining the “alleged height from which the shots were fired,” a conclusion his data could not yield. I began to wonder if his assessment of the flashlight, too, was faulty when he asserted that the blood on the lens was “back spatter” from a close-range shooting.

On the last day of class, I was given my “certificate of training” after receiving a 97 on my final exam. Everyone in my class passed. Griffin had told us that even if we failed the final, we would still receive a certificate of completion, but rarely, he added, did anyone fail. Our scores on our final exams were not recorded, he assured us, nor were the exams preserved. “Don’t worry that an attorney is going to come back and say, ‘You missed Question 14,’ ” he explained.

From time to time that week, Griffin cautioned us: “You won’t be walking out of here an expert. You’ll know just enough to be dangerous.” It was a startling statement, because judges across the nation have allowed police officers with no more training than we received — 40 hours — to testify as experts. Griffin reminded us that our class was merely an introduction to bloodstain-pattern analysis, and that we would need to complete an advanced class and a mentorship program before we would be proficient enough to call ourselves experts. Yet he advised us on what to say if we were called to testify in court. On the stand, he suggested, we should avoid saying what “probably” happened, because that would give an attorney who cross-examined us an opening. “You’ll be asked: ‘How probable? Eighty-five percent? Seventy-five percent?’ And you can’t say,” he told us, alluding to the fact that an analyst’s theory of a crime often cannot be substantiated with hard numbers. It was less risky, he said, to state, “The best explanation is. …”

III.

MY EXPERIENCE IN YUKON left me with more questions than answers, prompting me to reach out to two respected forensic scientists, Peter De Forest and Ralph Ristenbatt. De Forest taught criminalistics for 39 years at the John Jay College of Criminal Justice and studied in the 1960s under Paul Leland Kirk, a biochemist and forensic scientist who established a pioneering academic program in criminalistics at the University of California, Berkeley, in the 1950s. Ristenbatt teaches forensic science at Penn State University and is a former student of De Forest’s. Both men have been sharply critical of the phenomenon of the 40-hour class and of the “huge leap,” De Forest told me, that police officers who take such classes, and who lack a rigorous scientific education, often make when they use bloodstains to create a wholesale reconstruction of a crime. Curious what he and Ristenbatt would make of Thorman’s findings, I provided them with some basic information about the Bryan case, including the autopsy report, Thorman’s report, testimony and copies of photos of the flashlight and crime scene. Thorman himself had not examined the flashlight in person; he based his findings on photos alone.

De Forest and Ristenbatt were troubled by the significance that the prosecution ascribed to the flashlight, which Ristenbatt noted was “an isolated piece of evidence, found in isolation and without context.” Because it was not recovered at the crime scene, De Forest added, “its history is completely unknown. We don’t know when the blood got onto it or when it was placed in the trunk.” Moreover, they did not support Thorman’s conclusion that the tiny flecks of blood on the lens were back spatter. It was “irresponsible” to make such a classification, De Forest told me, given the small amount of blood involved and the limited surface area of the flashlight lens. “It’s totally specious, and there’s no evidence to support it,” he added. Ristenbatt explained that there were “many mechanisms other than gunshots that can cause similar patterns” but that it would be difficult to determine how the blood was deposited on the flashlight when it was divorced from the location where the actual bloodletting occurred.

During the course of our conversation, De Forest and Ristenbatt dismantled virtually every aspect of Thorman’s testimony. They rejected the notion that back spatter would not travel more than 46 inches, as Thorman testified. (Bevel’s textbook makes a similar claim, stating that small droplets can travel no farther than around 48 inches horizontally.) The distance airborne blood can travel, De Forest explained, is highly variable. He and Ristenbatt agreed that it was impossible to deduce from the available evidence that the killer had held the flashlight in his hand as he fired the fatal shots or changed his clothes in the master bathroom, as Thorman testified, before fleeing the house. Arriving at those conclusions “takes a lot of imagination,” De Forest noted. (Thorman declined to be interviewed for this article.) That Thorman had extrapolated far beyond what the evidence supported was the natural outcome, Ristenbatt told me, of a 40-hour class. “If you don’t understand the basic science, then you won’t understand its limitations,” he said.

Until the 1970s, the study of bloodstains was the sole province of those who did understand its limitations: forensic scientists. Working in the 1950s and ’60s, researchers, often in laboratories, observed the shapes and patterns blood created when it struck surfaces at various angles and velocities, and they sought to use that knowledge to better understand crime scenes. Many took their inspiration from Kirk, the forensic scientist De Forest studied under. Kirk was a brilliant scientific mind who had worked on the Manhattan Project. In 1966, he put bloodstain-pattern analysis on the map when he testified for the defense in the sensational retrial of Sam Sheppard, an Ohio doctor who served nearly a decade in prison for the murder of his pregnant wife. Sheppard always insisted that he tried to fight off an intruder on the night his wife was killed. At Sheppard’s retrial, Kirk laid out a detailed analysis of the blood spatter in the couple’s bedroom, which lent credence to the defendant’s account. Sheppard was subsequently acquitted, and Kirk’s forensic analysis was credited with almost single-handedly reversing the outcome of the era’s most notorious murder case.

But this technique did not stay in the lab for long. In 1973, Herbert Leon MacDonell, a research chemist and criminalist in Corning, N.Y., decided to take the complex concepts Kirk applied in the Sheppard case — namely, fluid dynamics and high-level math — and make them accessible to law enforcement. He did so by teaching what he called weeklong “institutes” at Police Departments around the country. From Springfield, Ill., to Tampa, Fla., to San Francisco, Police Departments welcomed the idiosyncratic MacDonell, who styled himself a modern-day Sherlock Holmes, even posing, on the cover of his book, “After Holmes,” with a pipe and the fictional sleuth’s trademark deerstalker hat. He believed investigators could use the droplets, spatters and trails of blood at crime scenes “to reconstruct the conditions at the moment of bloodshed,” as he asserted in an early report on the subject, “Flight Characteristics and Stain Patterns of Human Blood.” This mirrored what Kirk tried to do in the Sheppard case. As MacDonell spread the gospel of bloodstain interpretation, he instilled in police officers — some of whom had only a high school education — the belief that they, too, could unlock the secrets of a crime scene, so long as they possessed keen observational skills, a scientific calculator and the requisite course fees, which ranged, over the years, from roughly $200 to $700.

Of the more than 1,000 people MacDonell taught, some went on to lead bloodstain classes themselves, passing on his teachings to an ever-widening circle of police officers, investigators and forensic examiners. Many of these self-styled bloodstain interpreters testified as expert witnesses in trials, citing the classes they took as evidence of their proficiency. MacDonell himself testified around the country for four decades, helping to build bloodstain-pattern analysis’ reputation and legitimacy.

Some defense attorneys, whose clients were convicted on the basis of MacDonell’s testimony, began questioning whether this forensic discipline was as reliable as he claimed and mounted appeals challenging its validity and his credentials. Yet appellate courts in state after state upheld his right to testify as an expert. In 1980, when the Iowa Supreme Court affirmed the use of bloodstain-pattern analysis in State v. Hall, a case in which a man was convicted of stabbing his girlfriend to death, it explained its reasoning by citing MacDonell’s extensive experience, his standing as the field’s leading expert, the existence of bloodstain-pattern-analysis courses at several major schools and its use by Police Departments around the country. These were all credentials that traced back to MacDonell.

Rather than serve as gatekeepers who assessed the credibility of supposed scientific testimony before admitting it into evidence, many judges simply depended on the previous use of bloodstain-pattern analysis in the courts when deciding whether to allow it. Legal precedent, not science, served as their guide.

MacDonell continued to be a sought-after expert witness. In 1995, he testified for the defense at the O.J. Simpson trial, arguing that a sock found on Simpson’s bedroom floor, which was stained with the blood of his murdered ex-wife, Nicole Brown Simpson, was not the irrefutable proof of guilt that it appeared to be. The blood had not spattered onto the sock at the scene of the crime but rather had been transferred onto the sock by someone’s hand or an object, he told the jury, advancing the defense’s claim that Simpson was not a killer but the victim of a police frame-up. When grilled about his findings on cross-examination, MacDonell acknowledged that he was offering “an educated guess based upon experience.”

This is true not only for MacDonell but also for many bloodstain-pattern analysts. Despite the claims that some make on the stand, bloodstains can reveal only so much. Their shape, dimension, location and distribution may offer basic clues about what occurred — a trail of blood leading away from a victim, for example, might indicate that the perpetrator was injured when he fled — and may sometimes allow analysts to calculate where the bloodletting originated. But bloodstains are just one discrete part of the evidence left behind at a crime scene. Practitioners who rely only on such fragmentary pieces of information to reconstruct crimes are often engaging in nothing more than guesswork. Ultimately they can offer a theory, as MacDonell did in the Simpson trial. It is not unusual, in fact, for bloodstain-pattern analysts to face off in court, having drawn diametrically opposite conclusions. In one trial I attended, the prosecution’s expert saw proof of murder, and the defense’s expert saw suicide. Each pointed to the very same bloodstains as evidence.

. . . 

In 2009, the National Academy of Sciences sought to answer those questions with a groundbreaking, and damning, report. Its authors found that many forensic disciplines — including the analysis of blood spatter, hairs, bite marks, shoe and tire impressions and handwriting — were not as scientific as they often purported to be. Rather than being firmly grounded in hard data and rigorous, peer-reviewed research, many of these disciplines relied on the individual judgments of practitioners. The report included a sobering appraisal of bloodstain interpretation. Analysts’ opinions were often “more subjective than scientific,” its authors warned, and open to “context bias.” They noted that “some experts extrapolate far beyond what can be supported.” Moreover, “the complex patterns that fluids make when exiting wounds are highly variable,” they observed, and “in many cases their interpretations are difficult or impossible.” In conclusion, the authors cautioned, “the uncertainties associated with bloodstain-pattern analysis are enormous.”

Such uncertainties can have catastrophic real-life consequences for those accused of crimes. In one notorious case in 2000, David Camm, a former Indiana state trooper, told law enforcement that he had come home from playing basketball to find his wife and two children shot to death in their garage. Despite the fact that numerous witnesses placed Camm at the basketball game around the time of the murders, investigators believed that he was the killer. Over the course of three trials, prosecutors presented a succession of bloodstain-pattern analysts who testified that eight specks of blood found on the T-shirt he wore that night were “high-velocity impact spatter,” which suggested that he was present at the time of the shooting. The defense produced its own bloodstain experts, who argued that the specks in question were actually “transfer stains” — blood that blotted Camm’s T-shirt after he returned home and tried to render aid. (This divergence in expert opinion, one witness noted, amounted to “a 50 percent error rate.”) After spending 13 years behind bars, Camm was acquitted in 2013. Another man, a burglar with a long rap sheet whose DNA was found at the crime scene, was convicted of the murders and remains in prison.

Camm’s is not the only troubling case in which prosecutors have used bloodstain-pattern analysis to convict people whom the courts later freed. From Oregon to Texas to New York, convictions that hinged on the testimony of a bloodstain-pattern analyst have been overturned and the defendants acquitted or the charges dropped. As recently as this February, a judge vacated the conviction of a Missouri man named Brad Jennings for the 2006 murder of his wife, Lisa. After Jennings’s trial, evidence emerged that supported his story that his wife committed suicide.

Yet nearly a decade after the National Academy of Sciences report, little new work has been done to establish whether bloodstain-pattern analysis is actually a reliable forensic discipline. Few peer-reviewed studies exist, and research that might determine the accuracy of analysts’ findings is close to nonexistent. Meanwhile, experts with limited training continue to testify. While there is no national database of expert witnesses, or an index that lists cases in which bloodstain-pattern analysis played a role, I was able to search state appellate court rulings for cases in which bloodstain analysis was considered during the appeal. Though such a sampling represents a small fraction of cases that move through the criminal-justice system, the results were startling. These appellate rulings cited the testimony of investigators with 40 hours of training — and in several cases, less than that — in trials in Arizona, Florida, Idaho, Indiana, Kansas, Louisiana, Mississippi, Montana, New Jersey, Ohio, South Dakota and Texas. It is fair to assume that there are more defendants like David Camm out there.

VI.

“NEEDLESS TO SAY — I am devastated about not making parole,” Joe wrote to Leon Smith on Oct. 9, 2010, less than a week shy of the 25th anniversary of Mickey’s murder. “I was so upset that for several days I just shook with anger.” Joe’s 70th birthday had just passed, and he was slower on his feet; he had developed a heart condition and circulatory problems, and that fall, he was in poor health. “I’ve been back and forth to John Sealy Hospital so much in the last two months that my mail did not catch up with me until yesterday,” he wrote, referring to the Galveston hospital where he was transported for medical care. “I did get a new battery for my pacemaker, so I don’t have that worry.” Before he signed off, he expressed his gratitude to Smith. “Thank you for believing in me,” he wrote. “Give my best regards to Carole.”

It was the second time the Texas Board of Pardons and Paroles had rejected Joe’s bid for freedom. He became eligible for parole in 2007, after serving 20 years in prison. Despite the gravity of the crime he stood convicted of, he had hoped his pristine disciplinary record, unsullied by even a minor infraction, made him a promising candidate for early release. He had distinguished himself within the Walls, where he served as the pianist for the prison’s weekly chapel services, and he had collected enthusiastic letters of support, including a rare endorsement from his job supervisor, who told the parole board that Joe was “an exemplary model offender,” adding: “Everyone echoes the same opinion of Joe Bryan. He does not belong here.” Yet because the board’s deliberations are kept secret, Joe would never learn why his requests for parole were denied, or who might be protesting his release, or what their letters of opposition asserted. Nor would he know if not expressing remorse — he told board members he had been wrongly convicted — was considered a mark against him.

In 2010, as Joe’s odds of winning parole looked increasingly slim, he received some encouraging news. One of his first cousins, a court reporter, had asked a Waco attorney named Walter Reaves to look at Joe’s case years earlier, and Reaves had recently decided to take the case to Baylor Law School, where he supervised a clinic focusing on wrongful convictions. Hard-working and professorial, Reaves was known for taking on long-shot cases, particularly those in which flawed forensic testimony helped secure a guilty verdict. In 2001, he won the freedom of a Waco man named Calvin Washington, who was convicted of a 1986 murder partly on the dubious testimony of a later-discredited forensic dentist, who used a single bite mark on the victim to link Washington’s co-defendant (and, by implication, Washington) to the crime. The victim had been sexually assaulted, and DNA testing of the rape kit later excluded both men. Reaves also handled the unsuccessful appeal of Cameron Todd Willingham, an East Texas man who was executed in 2004 for killing his three children in a house fire — a conviction that rested on the forensic analysis of arson investigators whose methods were later called into question.

Reaves was troubled by the circumstantial nature of the Bryan case, from the paucity of evidence that placed Joe in Clifton at the time of the murder to the lack of scientific rigor in Thorman’s analysis. Thorman’s testimony at both trials, Reaves told me, was “nothing short of appalling.” Still, he knew the chances of getting Joe’s conviction overturned were slim. An appeals court had already rejected the notion that there was insufficient evidence to convict him. There was only one option left: to file a writ of habeas corpus, the final opportunity a defendant has, after all other appeals have been exhausted, to persuade the courts to review the case. To be successful, Reaves would have to produce new evidence that cast the reliability of the verdict into doubt.

Reaves decided to petition the court for DNA testing, though he doubted it would deliver the magic bullet the case needed. The circumstances of Mickey’s murder differed from those of many DNA exonerations, like Calvin Washington’s, in which a sexual assault occurred and a defendant could be excluded once a rape kit was analyzed. Because the murder happened in the Bryan home, Joe’s DNA would most likely be present, making an exclusion impossible. But there was always a chance that biological material was present that could, perhaps, lead to an alternate suspect.

In 2011, Reaves petitioned the court for DNA analysis of the cigarette butt, the flashlight and semen-stained underwear that was found in the wastebasket of the Bryans’ bathroom. The Bosque County district attorney’s office did not stand in the way of his request, but the results, which Reaves received in the summer of 2012, yielded no new information. No DNA profile could be obtained from the cigarette butt or the underwear. A partial profile on the lens of the flashlight “was too limited for a meaningful interpretation,” the report noted.

But amid the report’s technical language, one detail stood out — a single sentence about the flashlight, which stated: “A presumptive test for blood was negative on the lens.” In other words, the test could not confirm that what looked like blood spatter was actually blood. The report stopped short of determining why blood might not now appear to be present, but the bewildering results served as a reminder that Joe’s conviction hung by a gossamer thread. Twenty-seven years after the murder — despite advances in DNA testing that had unlocked the secrets to innumerable, seemingly irresolvable crimes — no one was any closer to knowing whether the minute, reddish-brown flecks on the flashlight’s lens were actually Mickey’s blood, or even whether they were blood at all.

Reaves was buoyed when a second-year Baylor Law student named Jessica Freud enrolled in his legal clinic in 2013, bringing a new energy and a boundless sense of optimism. Born in December 1985, the preternaturally cheerful Florida native was two months younger than the case itself. She had signed up for the clinic to see if she had an affinity for criminal-defense work, and when she finished reading the transcript of Joe’s retrial, more than 2,000 pages long, she had found her calling. “I couldn’t believe someone had been convicted on so little,” she told me. Not yet defeated by the recalcitrance of the criminal-justice system or the laboriousness of trying to win a new trial for someone twice convicted, she began brainstorming with her classmates: What investigative avenues, if any, had not been explored? Who was still alive? “Jessi gave me hope that we’d get somewhere,” Reaves said.

Over the next three years — first as a law student and then as a practicing lawyer with an office down the hall from Reaves’s — Freud collaborated with her mentor on Joe’s writ of habeas corpus. Smith, who had stepped down as editor of The Record, worked closely with the two lawyers as they looked for any information that might strengthen the appeal. He was elated when Reaves succeeded in getting the Clifton Police Department to grant access to the Bryan and Whitley files — everything that remained out of his grasp more than two decades earlier. Buried inside the banker’s boxes were recordings and transcripts of the interviews law enforcement conducted during its posthumous investigation of Dennis Dunlap. It was while poring over these files that Smith discovered a revelatory passage in an interview with one of Dunlap’s ex-wives, who told investigators of his terrifying, unpredictable behavior. He shot her children’s rabbits to death, she said, and once threatened to disfigure her; he also told her a few chillingly specific details about the Whitley murder, years after the fact.

When asked whether her ex-husband had ever spoken of other homicides, she told investigators that he had bragged, improbably, of having had an affair with Mickey Bryan. He also claimed to have been with her shortly before she was killed. “All he told me was that he dated her,” Dunlap’s ex-wife stated. “He dropped her off that night or that evening, and they had said they were going to break up.” Neither investigator had asked her a single follow-up question, quickly returning to the subject of Whitley. Despite the efforts of Smith, Freud and a private investigator, no one succeeded in getting the ex-wife to respond. But Freud learned from Susan Kleine, the fifth-grade teacher who had worked across the hall from Mickey, and Linda Liardon, a good friend of the Bryans’, that Dunlap separately made unwanted overtures toward them when he was on the Clifton police force — leering encounters that had spooked them so much that as each woman drove away from him, she took a meandering route home to throw him off her trail. In Kleine’s case, Dunlap briefly followed her before disappearing into the night.

Was it possible that Mickey and Dunlap crossed paths in the hours leading up to her death? Freud found another clue in the records of Ranger Joe Wilie, the lead investigator: On Oct. 16, 1985, the day after the murder, Wilie jotted down a note about two women — “Met Dunlap at Lazy Fisherman Friday nite?” — who may have been with Dunlap at a restaurant in Waco. Though Dunlap had left Clifton soon after Whitley was killed, before Mickey’s murder, the note suggested that he might have returned to the area around the time of Mickey’s death. This lead did not appear to have ever been followed up on.

Instead, the full force of the state’s resources had been brought to bear on Joe, even as prosecutors had searched for a theory about why he would have murdered his wife. In a memo that Reaves and Freud found in the district attorney’s files, under the words “Possible Motives,” someone had typed: “Need to get out of second marriage without a divorce.” “Another romantic interest.” “Money.” “Wife starting menopause or current bitch.” The No.1 theory, which topped the list, was a single word: “Queer.”

In their writ, which they filed in 2016 and amended the following year, Reaves and Freud advanced numerous claims to support their argument that Joe was entitled to another trial, contending that the newly discovered evidence about Dunlap could have swayed the jurors. They focused particular attention on the role that bloodstain-pattern analysis had played in winning the two convictions. Challenging the soundness of the conclusions Thorman made on the stand, Reaves and Freud pointed to his lack of experience in the field and the inadequacy of his training. “Much of his testimony was contrary to known and accepted science, or exhibited a lack of understanding of the relevant scientific principles,” the two lawyers wrote. Allowing someone with such limited proficiency to testify as an expert, they argued, was the equivalent of allowing a first-year law student to represent a defendant in a capital-murder case.

VII.

LAST SUMMER, Reaves and Freud filed a motion, separate from the writ, requesting that the remaining forensic evidence in the case undergo DNA testing. This included a human hair found in the trunk of Joe’s Mercury that did not match either of the Bryans, along with fingernail clippings and vaginal swabs taken at Mickey’s autopsy. Though the evidence was old and of uncertain significance — no semen was detected on the swabs back in 1985 — rapid advances in DNA analysis held out the promise, however remote, that new information might be gleaned. Reaves and Freud also sought further examination of the flashlight, in the hope of understanding why, in 2012, it tested negative for the presence of blood.

But under the newly elected Bosque County district attorney, Adam Sibley, their request met with unexpected resistance. Sibley’s predecessor, B.J. Shepherd, cooperated with Reaves’s original request for testing, but for reasons that have never been made clear, Sibley has taken a harder stance. (Sibley did not respond to multiple requests for an interview; nor did Andy McMullen, who prosecuted the Bryan case and who left the district attorney’s office in 1997.) Joe was not entitled to further testing under Texas law, prosecutors argued in court filings, because analysis of the items could not exonerate him; at best, it could only cast doubt on his conviction by suggesting the presence of a possible alternative suspect. “A defendant,” Shaun Carpenter, an assistant district attorney, wrote, “is not entitled to testing that merely muddies the waters.”

Judge James Morgan, who presided over both of Joe’s trials three decades earlier, appeared skeptical of the state’s logic at a hearing last August, pointing out that the use of DNA analysis in criminal investigations came about only after Mickey’s murder. “What’s the harm?” he pressed. “This wasn’t available to us back in ’85.” He later issued an opinion ordering testing to proceed. Still, the district attorney’s office did not back down. Prosecutors appealed the decision to Texas’ 11th Court of Appeals, which has not yet handed down an opinion. Should the court rule that DNA testing may move forward, prosecutors will most likely appeal the decision to the Texas Court of Criminal Appeals, the state’s highest criminal court, a move that could effectively delay testing for years. “The million-dollar question is, Why is the D.A.’s office fighting this so hard?” Reaves told me.

In late 2016, Reaves got a welcome bit of good news. The Texas Forensic Science Commission, which investigates complaints about the misuse of forensic testimony and evidence, announced that it had agreed to review another murder case that hinged on a bloodstain analyst’s testimony. Created by the State Legislature in 2005 in the wake of a scandal at the Houston Police Department’s crime lab, the commission — made up of seven scientists, one prosecutor and one defense attorney — does not investigate defendants’ guilt or innocence but rather the reliability and integrity of the science used to win their convictions. Its investigations often grow beyond the individual cases before them, extending to entire forensic fields, and it has emerged over the past decade as one of the most influential bodies in the country in advancing reforms. After the National Academy of Sciences issued its 2009 report, the commission made sweeping recommendations. It issued detailed findings about how to modernize arson science and improve the interpretation of DNA mixtures, and in 2014, it began the first review in the nation of state convictions based on microscopic hair analysis, a widely used technique whose accuracy has been challenged. Two years later, the commission called for a moratorium on the use of bite-mark evidence, having been unable to validate a basic premise of forensic dentistry: that a suspect can be identified from a bite mark on a victim’s body. But in its 11 years, the commission had yet to scrutinize bloodstain-pattern analysis.

The troubling case that prompted its inquiry centered on the 1987 murder of a man named Ed Clark, who was shot and killed in bed. His wife, Norma, claimed to be asleep in another room at the time, and though she was investigated, a grand jury declined to bring charges. But in 2010, Houston cold-case investigators revisited the evidence and observed tiny stains on the nightgown Norma wore the night of the murder. Certain they were looking at high-velocity impact spatter, which indicated that she was in proximity to her husband when he was shot, Norma was charged with murder and extradited from Tennessee, where she had lived quietly for years. Yet all the stains but one — a single microscopic spot — came back negative for the presence of blood. No DNA profile could be established from the speck, and whether it was her husband’s blood remained unknown. Despite a previous assault on Ed in the Clarks’ home by an unknown attacker and a death threat left on the couple’s answering machine days before Ed’s murder, Norma Clark was convicted in 2013.

Shortly after the commission’s announcement that it was looking into the Clark case, Reaves filed a complaint asking its members to also review Joe’s case. “Thorman was not qualified then, or now, to testify, and his conclusions are not, and were not, supported by valid scientific principles,” Reaves wrote. “The standards for admitting such testimony should be far more rigorous than simply attending one 40-hour course.” (Thorman told me that he couldn’t comment until after the commission’s investigation was completed.)

Joe’s lawyers were thrilled when the commission members decided that his case was worthy of further investigation. “The question on everyone’s mind was, How could someone with such limited experience testify on something so significant?” Lynn Garcia, the commission’s general counsel, told me. “Were there any standards that governed who could testify as a bloodstain expert?” With those questions in mind, the commission decided not only to investigate the integrity of the bloodstain interpretation in Joe’s case, but also to go one step further and examine the way in which bloodstain-pattern analysts are trained, with an eye toward setting standards for a discipline that had, for decades, managed to resist them.

This past Jan. 22, the commission convened a first-of-its-kind hearing on the subject and invited leaders in the field, including forensic scientists like Peter De Forest, high-profile bloodstain-pattern analysts like Tom Bevel and representatives of the Texas Rangers and other law-enforcement agencies. The group met in Austin in a modest conference room at the Supreme Court of Texas. De Forest spoke passionately of the need for analysts to have a formal scientific education, but many participants bristled at the idea. Troy Wilson of the Texas Rangers told the group that he didn’t think a bachelor of science degree made anyone more qualified to perform bloodstain-pattern analysis. “It means I can show up for class,” he said, “it means I can take a test and it means I can graduate.” Toby Wolson, a criminalist formerly with the Miami-Dade Police Department, argued that training and experience were sufficient. “Thomas Edison was self-taught,” he added. (Bevel declined to attend but wrote to the commission to recommend that practitioners take more 40-hour classes — specifically, two introductory and two advanced courses with different instructors.) The analysts’ resistance to change did not sit well with Jarvis Parsons, the commission’s sole prosecutor. “We are talking about the liberty of individuals,” he interjected at one point.

In February, the commission made one of the most consequential decisions in the field’s short history. It stipulated that bloodstain-pattern analysis must be performed by an accredited organization if it is to be allowed in court. Accreditation is a labor-intensive process and may be too cumbersome for individuals or small consulting firms. Someone like Thorman, who had not been affiliated with a crime lab, would have difficulty complying with the new standards. “Analysts will have to undergo proficiency testing,” Garcia told me. “Their cases will be reviewed, and there will be an outside audit of their work each year. Their testimony will be monitored to ensure they aren’t overstating their findings in court.” Many details, like educational requirements, have yet to be worked out; the change does not take effect until May 2019. Though the commission’s decision affects only Texas, it is expected to prompt other states to follow suit, as its reforms often do.

It was a huge leap forward, though the commission never took the further-reaching step of examining whether bloodstain-pattern analysis as a whole was scientifically valid. That question lay outside the group’s scope, Garcia explained. Unlike bite-mark evidence, which the commission had issued a moratorium against, bloodstain-pattern analysis has a national accrediting process (though many practitioners operate independently and without accreditation), so any investigation of its reliability, she explained, is a matter for the courts. This presents a Catch–22; it is the judiciary that has the authority to explore whether bloodstain-pattern analysis is scientifically valid, but it is judges who have proved to be the most reluctant to examine the soundness of forensic science. Indeed, despite the commission’s work showing that some forensic disciplines are far from scientific, judges have been slow to act; rather than rooting out dubious specialties like bite-mark evidence, they have continued admitting it in criminal prosecutions.

. . .