On her second day of testimony at the U.S. District Court in Tucson, Arizona, retired Pima County Sheriff’s Sgt. Sonia Pesqueira squinted at the monitor on the witness stand. It showed an item found at a crime scene in a case she investigated back in 1994: the murder of a 4-year-old girl. The item — a letter recovered from a trailer where the child had lived — was just one piece of evidence that could have provided an early lead.

Assistant Arizona Federal Public Defender Cary Sandman asked Pesqueira if she recognized the document. She peered at the monitor. “I can’t see what it says,” she said. She had left her glasses at home.

“Can I stop you for a moment?” U.S. District Judge Timothy Burgess interrupted. It was not yet 10 a.m. The evidentiary hearing had been on the docket for months. He had traveled from Alaska to preside. “It is not acceptable for us to have planned for so long for this hearing and have the witness now tell us that she can’t read these documents,” he said.

Pesqueira apologized. But patience was in short supply. Her investigation had sent a man to death row. For 23 years, that man had sworn he was innocent. There were good reasons to think he might be telling the truth, but prosecutors had fought to preserve his conviction. Now, revisiting her role as lead investigator, Pesqueira gave vexing testimony. There were just too many clues she had apparently missed; conspicuous leads she had failed to see. The forgotten glasses were almost too obvious a metaphor.

Burgess asked Pesqueira how far away she lived. Thirty minutes, she said. Were her glasses over the counter? Yes. Burgess handed her his own glasses. She put them on and smiled. “That helps a lot,” she said.

Across the courtroom, Barry Jones sat in an orange jumpsuit, shackles, and a belly chain. He had not seen Pesquiera in over two decades. But he remembered her well. In 1994, she had mercilessly interrogated Jones at the Pima County Sheriff’s Department, accusing him of killing his girlfriend’s 4-year-old daughter, Rachel Gray. Jones had taken the little girl to the hospital hours before, but she was dead on arrival. She had a scalp laceration and was covered in bruises. She was also bleeding from her vagina. Before the medical examiner had determined a cause of death, Jones became Pesquiera’s sole suspect. In 1995, he was convicted of rape and murder and sentenced to die.

With his appeals exhausted, Jones got a lucky break in 2012: a U.S. Supreme Court ruling that would allow him back into federal court on the basis of ineffective assistance of counsel. Jones’s defense had been profoundly flawed. His trial lawyers failed to call a single expert witness, putting only his 12-year-old daughter on the stand. In 2017, Burgess granted Jones a rare evidentiary hearing to consider whether his defense violated his Sixth Amendment rights. If Jones prevailed, he could win a new trial — or even be released.

The seven-day hearing began in late October 2017, at the Evo A. DeConcini U.S. Courthouse, a six-story postmodern building on the edge of downtown Tucson. Among the first witnesses were Jones’s original attorneys, Sean Bruner and Leslie Bowman, who admitted to numerous failures in representing him at trial. They were followed by a procession of experts across forensic disciplines: pathology, biomechanics, psychology, and bloodstain interpretation. They gave compelling testimony that jurors had never heard — evidence that would likely have raised doubts about Jones’s guilt.

Burgess was attentive, asking lots of questions. Pesquiera was the penultimate witness. In a black top and bright blue cardigan, she carried the confidence of a veteran cop. Prosecutors from the Arizona Attorney General’s Office had asked that she be allowed to sit at their table throughout the hearing, a somewhat unusual request that signaled her enduring commitment to the case. But now, on the stand, Pesqueira seemed less self-assured — sometimes impossibly obtuse. When asked about a specific lead she never pursued, she conceded it was absent from her reports. “Which also means if it’s not in your report, you didn’t look at it?” Burgess asked. “I didn’t document it, yes,” Pesqueira repeated. There’s a difference between documenting and doing, the judge explained. “If you did something, you’d document it, right?” Yes, she said.

In other words, if it was not in the police report, it hadn’t been done.

The investigative records in the Jones case were riddled with holes. Most alarming, there was no sign that Pesqueira ever sought critical information about how, exactly, Rachel was killed. The child had died as a result of a sharp blow to the abdomen that perforated part of her small intestine, the duodenum. Leakage from the tear caused a deadly condition called peritonitis, which, if left untreated, can lead to septic shock. To estimate when the blow occurred, a pathologist testified, one would need to thoroughly investigate when Rachel showed her earliest symptoms of illness. Pesqueira never did.

Pesqueira-3-1517593756

Pima County Sheriff’s Detective Sonia Pesqueira interrogates Barry Jones in 1994.

Photo: Pima County Sheriff’s Department


Instead, she had drawn conclusions based on her own first impressions at the hospital. “I do my own examination,” she told the grand jury in 1994. “Medical personnel are not in the room.” Given Rachel’s appearance — and the estimate that she had died in the early hours of May 2, 1994 — Pesqueira concluded she had been fatally beaten the day before. This made Jones the likeliest suspect. In a deposition before the 2017 hearing, Pesqueira was asked if she had ever identified anyone who might have hurt Rachel in the days before May 1. “I had no reason to believe [her injury] had gone that long, so no,” she said.

If Pesqueira had expanded her investigation, she would have found plenty of alternative suspects. There was Rachel’s own mother, Angela Gray, for one. Unlike Jones, she had a reputation for abusing her kids. (“That would have been a good thing to have,” Pesqueira said at the hearing when shown a relative’s pretrial statement saying that Gray had hit her kids in the stomach and thrown them against the wall.) There was also a little boy whom Rachel had accused of hitting her in the stomach with a stick or an iron bar, according to kids at the trailer park. Pesqueira dismissed this as a rumor.

Yet another set of people might have been questioned about Rachel’s vaginal injury. The injury was likely much older than it appeared, doctors explained, possibly even predating the period of time during which Rachel lived with Jones. This introduced an array of other suspects. There was Rachel’s 14-year-old half-brother, who had a reputation for preying on girls at the trailer park. (The letter that prompted the exchange over the glasses was one clue about his sexual activity.) There was also her mother’s abusive ex-boyfriend, Zoly, with whom Rachel and her siblings had lived before moving in with Jones.

“I am trying to understand why there was no follow-up with Zoly,” Burgess told Pesqueira toward the end of her testimony. There was no evidence that Jones had ever bathed or dressed the kids in their brief time living with him. Was Zoly ever responsible for doing such things? “I don’t know that,” Pesqueira said.

“That would require interviewing him, right?” Burgess asked.

“Yes, sir,” she said.

Did she think this was important?

“Yes, sir.”

But Pesqueira could not explain why she had never spoken to Zoly. “I think I just — I didn’t do it.”

PastedGraphic-5-1517428021

The yellow van driven by Barry Jones, who worked as a mechanic. The inside was divided into compartments filled with tools.

Photo: Pima County Sheriff’s Department


A week before the hearing in Tucson, The Intercept published a lengthy investigation into the case of Barry Jones. Based on police reports, juror interviews, and thousands of pages of court records, it revealed several hallmarks of wrongful convictions, from junk forensics to prosecutorial misconduct. At the center, however, was a police investigation that succumbed to tunnel vision from the start.

Jones first came under suspicion after leaving Rachel and her mother at Tucson’s Kino Community Hospital shortly after 6 a.m. on May 2, 1994. Law enforcement found it strange that he did not return — and before long, so did Gray. As she would tell Pesqueira, she and her three children had only lived with Jones at his home in the Desert Vista Trailer Park for a month or so. Gray said she’d spent most of the previous day asleep; when she woke up, Jones told her Rachel had fallen out of his parked work van, bloodying her head. He said he had taken her to the nearby Rural Metro Fire Station for medical attention. But later, Rachel started vomiting. She was thirsty but could not keep liquids down. After putting her to sleep, Gray found Rachel unresponsive the next morning, with her eyes rolled back in her head.

Jones would later admit that he had lied about taking Rachel to the fire station. He explained that he’d seen a cop there and he did not want to be caught driving with a suspended license. So he drove on to a Quik Mart, where he spotted a paramedic. The lie — and the fact that Jones had been seen taking Rachel on short trips in his van on the afternoon of May 1 — became the early basis for the state’s case. At trial, prosecutors also leaned on the eyewitness accounts of 8-year-old twins, Ray and Laura Lopez, who said they had seen a man who matched Jones’s description hitting a young girl while driving a van.

But despite the circumstantial evidence, there was no physical evidence linking Jones to Rachel’s injuries apart from some small drops of blood found on his clothes and the passenger side of the van. Nor was there any evidence he had raped her. The most critical items for forensic testing — such as the clothes Rachel and Jones were wearing on the day he allegedly assaulted her — were never even collected. Other vital pieces of evidence were lost, concealed, or destroyed.

Responsibility for the shoddy police work fell on Pesqueira. But the conviction also rested on the trial testimony of Dr. John Howard, a pathologist with the Pima County Medical Examiner’s Office. Howard had testified that Rachel’s fatal abdominal injury was inflicted some 12 hours before her death — an estimate that fell neatly within the window during which the state alleged Jones had Rachel in his van. This was the linchpin in the state’s case against Jones. Yet — in a serious discrepancy that Jones’s lawyers failed to catch — Howard had given a strikingly different calculation mere weeks earlier, at Gray’s trial. (She was convicted of child abuse but acquitted of murder.) There, Howard had testified that the abdominal injury was “most consistent” with occurring 24 hours or more before Rachel died, long before she was with Jones.

Whatever the reason for Howard’s disparate statements in 1995, the state’s medical timeline has since been dismantled. Beginning in 2002, experts hired by Jones’s federal public defenders firmly concluded that Rachel’s fatal injury could not possibly have occurred within the window presented at trial. In a significant concession, the Arizona Attorney General’s Office wrote to Jones’s attorneys in advance of the 2017 hearing acknowledging that the current Pima County medical examiner “did not dispute the conclusions of your experts.” Finally, Howard himself had since shifted his stance in multiple statements. If attorneys had asked whether Rachel’s fatal abdominal injury could have happened more than 24 hours before her death, he explained in 2004, “I would have answered the question in the affirmative.”

Nevertheless, the state denied that its case against Jones had fallen apart. In an email to The Intercept last fall, a spokesperson for the attorney general’s office rejected the notion as “completely inaccurate.” With its medical theory debunked, it seemed uncertain how Arizona would defend its case at the hearing in Tucson.

A view of the outside of the Evo A. DeConcini Federal courthouse Wednesday, May 25, 2011 in Tucson, Ariz.  Suspected shooter Jared Loughner, who is charged with shooting U.S. Rep. Garbrielle Giffords, D-Ariz. and 18 others, is in court for a competency hearing Wednesday to determine if he is fit to stand trial.  (AP Photo/Chris Morrison)

A view of the Evo A. DeConcini U.S. Courthouse on May 25, 2011, in Tucson, Ariz.

Photo: Chris Morrison/AP


On the first day of the hearing, Assistant Attorney General Lacey Gard, who heads the capital unit for the Arizona solicitor general, made the answer clear: Prosecutors would double down on the original theory of the crime, never mind what medical experts, the current county medical examiner, or even Howard himself had said. “His opinions are the same now as they were at trial,” Gard declared in her opening statement.

Gard insisted the evidence still pointed to Jones, while reminding the court that the question of innocence was irrelevant: The matter at hand was Jones’s Sixth Amendment claim. Under the landmark U.S. Supreme Court case Strickland v. Washington, this meant Jones had to prove, first, that his original attorneys had been deficient, and second, that their failures were so severe that they affected the outcome of the case. But this second question was irrelevant, too, Gard argued. “There clearly was no deficiency.”

To prove this, Gard and her colleague Assistant Attorney General Myles Braccio pointed to a series of half-steps in the right direction. For example, Jones’s trial attorneys had contacted an independent pathologist, Dr. Philip Keen, back in 1994. They sent him Howard’s autopsy report, asked preliminary questions about the timing of Rachel’s injuries, and asked for a full review of the medical evidence. Yet there was no record that the additional materials were ever sent. There was proof of a short follow-up phone call, but no one actually remembered it. According to Braccio, the “only conclusion we can draw from these records” is that Keen had agreed with Howard. “Or that we dropped the ball and didn’t follow up properly,” Bowman, Jones’s former attorney, responded.

There were several such attempts by the state to recast the absence of evidence as a sign of a job well done. Braccio pointed to gaps in the lawyers’ billing records, suggesting they had done more work than the documents showed. This was a death penalty case after all, “and realizing these high stakes,” he asked Bowman, “you have no reason to believe, as you sit here today, that … you devoted anything other than your best efforts, correct?” When Bowman, who is now a magistrate judge, said she and Bruner had not put their best efforts forward, Braccio reminded her that, under Strickland, it did not actually matter what she thought now. It is an “objective standard,” not based in hindsight.

But perhaps most revealing was the state’s attempt to rationalize the failure to secure expert witnesses. Unlike Jones’s current federal public defenders, who can afford “very expensive experts,” Gard said, trial lawyers had to file motions to seek funding from the presiding judge. “They can’t control what the court is going to do, and that doesn’t make them ineffective if they don’t get the same amount of funding,” Gard said. Braccio built on the argument. Wasn’t it true, he asked Bowman, that the trial lawyers faced “significant funding obstacles” from the judge?

Both lawyers denied that funding for experts would have been a problem if they had tried harder to get it. Yet Jones’s post-conviction attorney disagreed. The culture throughout Arizona at the time “was that you didn’t get experts, you didn’t get investigators, you didn’t get transcripts,” he insisted, a fatalistic belief that seemed clearly not to have served his client. Outside the narrow legal framework of the hearing, the notion that Jones should remain on death row because the system itself was stacked against him was disturbing.

Although they are both in their late 30s, Braccio was nowhere near as seasoned a litigator as Gard. But he was not new to legal contests over whether someone should live or die. As a young paralegal in the wake of Ring v. Arizona — a seminal U.S. Supreme Court ruling that said juries, not judges, should decide whether to impose death sentences — Braccio worked on the case that prevented the decision from applying retroactively. Written by Justice Antonin Scalia, the 2004 ruling kept Jones on death row.

There was no disputing that the timing of Rachel’s injuries was central to the case — and that the state’s theory no longer held water.

With a cocky air, Braccio seemed eager to win. Yet his scattered and ham-fisted approach seemed to irritate Burgess. So did his habit of asking leading questions, prompting consistent objections sustained by the judge. At times, Burgess seemed simply confused. “I’m sorry. Counsel, you know, look, this isn’t a guessing game where the witness is supposed to figure out what the exhibit is,” he told Braccio at one point. When Braccio’s cross-examination went off the rails, Burgess did not hesitate to interject with questions of his own.

Burgess seemed especially interested in the medical testimony, for obvious reasons. There was simply no disputing that the timing of Rachel’s injuries was central to the case — and that the state’s theory no longer held water. Although there were some differences in their analysis, the medical experts called by Jones’s attorneys all deemed it impossible for Rachel to have been injured so close to her death. This included Keen, the same pathologist who had agreed in 1994 to review the autopsy materials but never received them. Keen had since reviewed images of the tissue slides and other items necessary to determining the timing of Rachel’s fatal injury. Contrary to Braccio’s speculation, Keen firmly concluded that the state’s timeline was wrong. “Rachel’s small bowel laceration was not inflicted on May 1, 1994,” he wrote in a 2017 affidavit, and “Jones’s jury was misled to believe otherwise.” Another doctor testified that there were “no reported cases in medical literature in which this type of injury has resulted in death in less than 48 hours.”

On November 1, Jones’s attorneys called Dr. Janice Ophoven, a nationally renowned pediatric forensic pathologist. Ophoven, who first reviewed the tissue slides and other autopsy materials in 2002, has remained adamant that the state’s case against Jones was fatally flawed. “There is no question that the child suffered injuries at the hands of another,” she wrote in 2010. But the evidence used to convict Jones was “absolutely incorrect.”

Ophoven estimated that Rachel’s fatal injury had taken place at least two or three days before she died. To explain, she used a diagram to locate the duodenum, the part of the small intestine where Rachel was fatally wounded. It lies in a “special space,” she explained, between the peritoneum, the lining of the abdominal cavity; and the retroperitoneum, the area behind it. This separation is difficult but critical to understand. Whatever ruptured Rachel’s duodenum, Ophoven said, the immediate effect — an inflammation that becomes more lethal as it spreads — was restricted to the retroperitoneum; it did not expand to the rest of the abdominal cavity until much later. As a result, the symptoms typically associated with an abdominal injury were significantly delayed. Rachel might have felt some discomfort and nausea, Ophoven said, but even doctors can misinterpret such symptoms, leaving the developing peritonitis undetected. Ophoven said she had seen cases in which children were seen by physicians who “pat them on the head and tell them it’s the flu, send them home, and they’re dead the next day.”

Rachel’s symptoms were consistent with this scenario, Ophoven explained. Her “unquenchable thirst” on the night before she died was a sign of dehydration — a precursor to irreversible shock that takes time to develop. An even more concrete indicator was a post-mortem analysis of fluid from Rachel’s eye. It showed “a derangement of body chemistries that takes considerable time,” Ophoven explained.

But the key to figuring out Rachel’s death, Ophoven said, was identifying the point when she started getting sick. A neighbor named Isobel Tafe had told Pesqueira that she saw Rachel looking gray and unwell on Saturday, April 30 — two days before her death. Ophoven considered this an important clue — “the gray color is kind of specific to this kind of process,” she testified. Yet Pesqueira had discarded the statement at the time. “I thought she was giving me the wrong day,” she recalled at the hearing. “The incident we were discussing would have occurred on Sunday.”

Cross-examining Ophoven, Gard suggested she was cherry-picking statements to suit her analysis. After all, Rachel’s older sister had said she looked fine on Sunday morning. Why accept Tafe’s observation while discounting 11-year-old Becky’s? “The evidence contradicts it,” Ophoven responded. Regardless, she was hamstrung by a lack of information. “I needed to look at the last time the child was well,” she said, “and that would have been what the investigation warranted had they not focused on Mr. Jones immediately.”

There was another critical piece to Ophoven’s testimony: an explanation for Rachel’s disturbing appearance at the hospital. Neither Jones nor Gray had recalled seeing so many bruises on Rachel the night before she died. At trial, Howard had expounded on the age of the bruises based on their color and appearance, suggesting Rachel had recently been pummeled. But Ophoven called such analysis “hogwash.” To accurately assess a bruise, she said, you have to examine the tissue beneath the skin. What’s more, her own work had taught her that when children arrive at the hospital in shock, “you’ll start to see marks appearing on their body,” along with spontaneous bleeding. This happens when the body is no longer able to clot blood, a harbinger of irreversible shock. While Ophoven believed that Rachel was a battered child, the bruises on her torso and abdomen — and the blood in her vagina — were not necessarily evidence of new injuries, she explained. They were signs of the “terminal event,” she said, the final stage in her death from peritonitis.

Judge Steven Mauer listens as Dr. Janice Ophoven, testifies at Clackamas County Circuit Court in Oregon City, Ore. on Thursday, July 9, 2009. Carl and Raylene Worthington _ who believe in treating illnesses with faith-healing _ are charged with manslaughter and criminal mistreatment in the March 2008 death of their daughter Ava. (AP Photo/Ross William Hamilton, Pool)

Dr. Janice Ophoven testifies at the Clackamas County Circuit Court in Oregon City, Ore., on July 9, 2009.

Photo: Ross William Hamilton/Pool/AP


Day after day as the hearing wore on, Barry Jones sat at the defense table on the right side of the courtroom. The shackles dug into his ankles and a stun device made his arm sore. At 59, he looked nothing like he did the last time he was free. His whitening hair was combed back over his ears. Every morning since the hearing began, he had been woken up before dawn in his death row cell, waiting for the prison van to take him from Florence to Tucson.

Burgess had granted a request to allow Jones a free hand so that he could take notes, a portion of which were shared with me by his lawyers. “Mostly I am just writing so I can ignore the three police behind me and focus on other things than what is being said,” Jones wrote in neat blue cursive. Listening to the details of the case was like reopening an old wound, he wrote. To calm his nerves, Jones would try to distract himself.

Jones was buoyed by the presence of a handful of supporters in the courtroom. They included his niece and other family members he had not expected to see. He was especially surprised to see his older brother, Otis, a retired law enforcement officer who showed up with his wife. The brothers had not seen each other in years. No one from Rachel’s family was there. But after doing eight years in prison for child abuse, Angela Gray wrote in a 2009 affidavit that she no longer believed Jones had killed her daughter.

Jones was floored by the amount of technology surrounding him in the courtroom. The microphones and monitors were a reminder of how much the world had changed. As he marveled at the luxury of using an ordinary pen, life on the outside felt daunting. “Everyone is emailing, punching keys on their laptop computer, or using cellphones with the ability to take pictures,” he wrote one day. “Would someone like me fit in?”

Jones was especially glad to see Sylvia Lett, his former federal public defender, who has been a lifeline for him. At her home last fall, Lett showed me cards Jones had sent her over the years and described his guarded optimism about the upcoming hearing. “I think that he goes through real periods of time where he doesn’t want to hope,” she said. It was often easier to defer to his lawyers than to remain invested in every twist and turn of his appeals, so he’d say, “I trust you guys. I’m going to put my faith in you. I don’t need to hear about it,” Lett said. “And I think that has maybe also helped him in a weird way.”

Jones had good reason to keep from getting his hopes up. Courts have a tendency to give defense lawyers the benefit of the doubt when it comes to ineffective assistance claims. In Strickland, the Supreme Court held that judges must show deference to lower courts and “indulge a strong presumption that defense counsel was within the wide range of reasonable professional assistance.” This included cases in which missing records made it hard to assess whether lawyers’ performances were part of an overall strategy, which would blunt any attempt at relief.

In his notes, Jones repeatedly expressed gratitude toward his legal team, while trying hard to keep his optimism in check. Even if he died in prison, he wrote, at least he was able to come this far. But other times, he dared to imagine what it would be like to live outside. With the holidays coming, the prison had sent out the list of items at the “Christmas store,” products available for purchase from a private company. He would start saving money to buy canned oysters and clams, coffee, and some Hostess pies, he wrote. “I wonder how much Hostess apple and cherry pies cost on the outside?”

GettyImages-2739974-1517352685

Dr. John Howard, chief medical examiner for Pierce County, Wash., testifies at the Virginia Beach Circuit Court on Nov. 18, 2003, in Virginia.

Photo: Dave Ellis-Pool/Getty Images

By the time the last witness took the stand at the evidentiary hearing, experts had thoroughly discredited the case against Jones. An accident reconstructionist had examined his old van and visited the site where the Lopez twins had claimed to see Jones hitting a little girl while driving. Considering their height, the motion of the vehicle, and the position of the little girl in the passenger seat, he said, “My conclusion overall was that they could not have seen what they said they saw.” A biomechanics expert echoed this view, describing the twins’ descriptions as “extremely questionable.” A forensic psychologist raised serious concern over the “excessive leading and suggestive questions” during the twins’ interview by police. And a bloodstain analyst cast doubt on the opinions of Pesqueira, who had testified at trial that the blood evidence indicated Jones had assaulted Rachel in the van despite lacking qualifications to back up such analysis.

In response, Braccio used a common courtroom tactic: asking how much they had been paid. It’s an easy way to cast an expert witness as a hired gun in front of a jury, but seemed less likely to impress a federal judge. More convincing were the questions that forced the experts to concede that there were too many unknown factors at play for their conclusions to be completely reliable. But once more, that got to the heart of the problem: the lack of investigation in 1994.

On November 7, 2017, Dr. John Howard took the stand — the state’s second and final witness. Thin, pale, and wearing glasses and a dark suit, Howard looked every bit the veteran pathologist. After providing an off-the-cuff history of medical examiners, he testified that he had conducted “well over” 8,000 autopsies over the course of his career, including hundreds on children under 5 years old. He has served as president of the National Association of Medical Examiners — the group that accredits forensic pathologists all over the country. Today he works for the Spokane County Medical Examiner in Washington state.

Despite his credentials, there were reasons to be concerned about Howard’s work beyond Jones’s case. Prior to his return to Tucson in 2017, the Washington Department of Health had launched an official investigation into Howard. There now were complaints pending in seven separate cases, where his autopsies had come under question, attracting the attention of the press. In one, he said a woman whose naked body was found bound in plastic and duct tape had died of a heart attack associated with alcoholism. In another, a body had been found cut in half and left in the woods — he said the cause of death was undetermined.

Whatever was going on in Spokane, in Tucson, it didn’t matter. Like the question of Jones’s innocence, that was a conversation for another day.

“You have testified now three times under oath on this, right? Would you expect that there be some consistency between those testimonies?”

The monitors in the gallery were darkened as Braccio led Howard through the graphic autopsy photos. He asked about Rachel’s injuries one at a time. There was the scalp laceration, which Jones had said occurred when Rachel had fallen out of his van. It was an acute injury, Howard said, likely having occurred within hours to a day of her death. Then there was the vaginal injury. Ophoven had testified it was likely “weeks old,” and Keen said he would not even put it in the same context as the abdominal injury — “It’s not in the death timeline.” Yet Howard said it showed no signs of being an old injury. “It could be a few hours, it could be a day.”

Finally, Braccio asked Howard about Rachel’s abdominal injury. “My opinion again is this is an acute injury, could be a few hours, typical of a day or the same day as death,” Howard said, reverting to his opinion at Jones’s trial. Braccio showed him his previous statements suggesting the injury might actually have preceded May 1, 1994. Did this change his opinion? No, Howard said. You cannot date an injury exactly, he explained. “It could be just a few hours, it could be 24 hours, it could potentially, or at least in theory, be longer.”

When Sandman stood to cross-examine Howard, he looked weary. The hearing had consumed his office for months. He had deep circles under his eyes. “You have testified now three times under oath on this, right?” Sandman asked, speaking quietly. Yes, Howard replied. “Would you expect that there be some consistency between those testimonies you’ve given on the timing of Rachel’s injures?”  It was all a matter of how the questions were worded, Howard explained.

Sandman showed Howard portions of his testimony from the Gray trial and compared them to his testimony at the Jones trial. Howard conceded they could be interpreted as different. “Wouldn’t it be important for you to explain why you are coming up with a different time frame when you testify in the Jones trial?” Burgess asked. Howard said maybe he had reviewed the case further after Gray’s trial and given it more thought. The answer did not impress Burgess. “You understand that in these trials there was a lot at stake, right?” he asked.

Yet it would not take long for Howard to change his testimony again. On redirect, Braccio asked Howard if, in fact, he believed the opinions had been the same at both trials. Yes, Howard said, only the wording was different. “Dr. Howard,” Sandman said moments later, “you just told the judge a couple of minutes ago that your testimony in the Gray trial and the Jones trial were different, weren’t they?”

It was confusing, even surreal. It certainly did not instill confidence in Howard’s opinions. If presented to a jury, Sandman told me after the hearing, “I think they would have to conclude that Dr. Howard doesn’t know anything about when any of this happened.”

Yet this was not necessarily the most troubling part of Howard’s testimony in Tucson. Before the hearing ended, Burgess asked about other autopsies Howard had performed on young kids who died of peritonitis, inquiring what portion involved death within 24 hours of injury. “I’d say most of them, as I recall, but I can’t be certain,” Howard responded. He added, “There are many recorded cases of that.”

This clashed wildly with the previous experts’ testimony. Burgess wanted to be clear about what he was hearing. “So your testimony is you think there are many recorded cases in which there has been a similar injury to the duodenum that resulted in death in 24 hours or less?” Yes, Howard answered.

Burgess told Howard what the court had previously heard, specifically that there were no reported cases in published medical literature in which the same type of injury had resulted in death in less than 48 hours. “Hmm,” Howard muttered. He was unaware of that. “You think you have personally seen instances that you’ve worked on in which this has happened?” Burgess asked. “Yes,” Howard said. “In less than 24 hours?” Yes, he repeated. And the literature would support this? Howard began to say that not all medical literature is published when Burgess interrupted. “I think words are important,” the judge said. He was not asking what might exist in a file somewhere. Howard said he was not prepared to answer that. He would have to review the literature himself.

In an email following the hearing, Howard shared the results of his subsequent online search, which he also sent to prosecutors. It had taken him less than three minutes to find literature to support his testimony, he wrote, pasting in three short paragraphs from a 1997 article in a British medical journal, which contained a short description of a 4-year-old girl with a fatally ruptured duodenum and other injuries. The girl had presented “a one day history of increasing abdominal pain,” it said. This line was in bold and italics. He included no other examples. He also misspelled Rachel Gray.

If Arizona prosecutors considered Howard’s finding convincing, they chose not to include it in their post-hearing brief. For her part, Ophoven found the excerpt “meaningless.” It lacked basic details, she said. “We have no idea when the actual blow occurred that caused the fatality. We don’t know if the cause of the death in the belly was peritonitis. … That case report doesn’t tell us anything.”

Lawyers on both sides filed their final briefs in January. An oral argument is scheduled for March 2.