WILMINGTON, N.C. – They are old men now, the doctor and the lawyer, ancient adversaries confronting each other one last time.

The doctor shuffles into the courtroom, his feet in socks and slippers, his ankles in chains. Once a swaggering bon vivant with a Maserati, a yacht and a playboy lifestyle, he is now dreadfully diminished, drained of color and of dignity, with a prison pallor and hair gone white, insulted by a prison haircut. He wears a washed-out dun prison-issue jumpsuit with big black letters: INMATE NEW HANOVER COUNTY. The only daub of color is the maroon accordion file he holds. It glistens. It has been so often repaired over the years that it’s more Scotch tape than cardboard. Inside are papers that document a lost life. For nearly half that life, the doctor, 68, has been in federal custody, largely because of the efforts of the government lawyer who sits across the room, at whom he will not look, not even once, in the seven days they will spend together here.

Forty-two years ago, Green Beret Army doctor Jeffrey MacDonald stabbed and bludgeoned to death his pregnant wife, Colette, and his two little girls . . . or he didn’t. One currently popular theory is that he didn’t. That’s the contention of the renowned film documentarian Errol Morris, director of ”The Thin Blue Line” and ”The Fog of War.” When he couldn’t find backers for a movie about the case, Morris wrote the newly released ”A Wilderness of Error,” which concludes that MacDonald was railroaded by incompetent cops and overzealous prosecutors, and very likely is no murderer. A New York Times book reviewer agreed, declaring that the book will leave readers ”85 percent certain” of MacDonald’s innocence. Also weighing in with an online review was, of all people, Sarah Palin. She loved the book, for reasons of her own.

This evidentiary hearing in September in Wilmington is part of what many believe to be MacDonald’s last chance at freedom. It’s probably the penultimate step in an appeals process that began after the doctor’s initial conviction in 1979 and has continued ever since, driven and financed by a steadfast band of supporters. Private investigators have been deployed; teams of attorneys have been hired and replaced; appellate actions have been filed, argued, dismissed.

This most recent one – the one that brings us here – involves newly surfaced reports that a woman, long dead, had confessed to participating in the crime but was bullied into silence by prosecutors. Also, that there is new DNA evidence said to support what has been MacDonald’s contention all along – that the murders were committed by drug-addled hippie intruders imitating the Charles Manson killings.

It’s all laid out in Morris’ book. The flamboyant filmmaker sits in the courtroom, beaming as witness after witness takes the stand to raise doubts about the official version of the crime. Morris scowls whenever cross-examination makes these claims seem less solid, which is often.

Many old, familiar faces are here, including Bob Stevenson, 73, the murdered woman’s surviving brother, who sits ramrod straight, with carefully modulated, white-knuckled fury, unapologetically declaring to reporters that his sacred mission in life is to watch Jeffrey MacDonald continue to suffer in custody, and then outlive him. There is also Kathryn MacDonald, the 51-year-old children’s drama teacher from Columbia, Md., whom MacDonald married 10 years ago, and with whom he has never been alone. Defiantly upbeat, she declares her husband ”the strongest, most honorable and wonderful person” she has ever known, and claims they share an intimacy that would be envied by many conventionally married couples.

There is no shortage of emotion in this case, still.

Finally, there is Brian Murtagh, the plumpish little government prosecutor on the other side of the courtroom, a resolutely boring man. His most distinguishing feature is the pair of suspenders he always wears, even at home. ”It’s not a fashion statement,” he says pre-emptively, as if disturbed by the very idea. ”I need ’em to keep my pants up.”

Among the five attorneys here, Murtagh alone has been on this case from the start; the woman who sits to his left, Assistant U.S. Attorney Leslie Cooley, was born the year of the murder trial, nine years after the crime.

As the years have passed and the appeals continued, several times reaching the U.S. Supreme Court, as witnesses have died, as evidence has literally desiccated in file drawers, as lawyers came and went – his two co-counsel at the trial were later disbarred for ethics violations unrelated to this case – through it all, Murtagh alone persisted. At 66, he is now technically retired, but the government brought him back especially for this appeal. Jeffrey MacDonald hates him.

”The little viper” is how MacDonald described Murtagh to the grand jury hearing his murder case. Later, to reporters, MacDonald called Murtagh ”a berserk person, a middle-level bureaucrat run amok . . . paid with taxpayer money to do this to me.”

When Morris’ ”A Wilderness of Error: The Trials of Jeffrey MacDonald” came out in September, Brian Murtagh sat in the study of the Oakton, Va., home he shares with Margaret, his wife of 43 years, and read it cover to cover, all 500-plus pages. He found it credulous, manipulative, a Swiss cheese of strategic omissions. To assert this, he typed out a rebuttal – a legal brief, double-spaced, 14 pages long, with Roman numerals and alphanumerically labeled paragraphs. It is not light reading. Morris, Murtagh writes, ”doesn’t explain how 60 pieces of the pajama top, including the ripped-off pocket bearing a contact stain in Colette’s blood, could be found in the master bedroom, as well as 30 seam threads. . . . ” Murtagh didn’t file this odd document anywhere. He didn’t release it to the media. It was mostly for himself.

Murtagh sounds exactly like a lawyer but carries himself exactly like a butler. You want to call him Jeeves. He’s punctilious, a bit formal, often greeting people with a courtly little bow. He views this whole case with an air of bemused exasperation, puzzled by its refusal to die. He knows his ”brief” would mostly confuse people. Only two people on Earth, he says, are really in a position to understand it – to understand what a flimsy, paltry, bankrupt case for innocence Errol Morris makes.

Brian Murtagh and . . . ?

Murtagh smiles grimly.

”Jeffrey MacDonald.”

If you are of a certain age, the story is familiar. In the early morning hours of Feb. 17, 1970, in an officer’s apartment at Fort Bragg, N.C., someone savagely attacked Colette MacDonald, 26, and her two daughters, Kimberly, 5, and Kristen, 2. The weapons were a kitchen knife, an ice pick and a piece of scrap lumber used as a club. Colette was hit so forcefully that both her arms were broken. Kimberly’s skull was split open. All three were stabbed as though in a frenzy, the wounds coming from all directions, but unerringly finding vital organs and vessels.

Military police arrived, summoned by a gasping phone call from Capt. Jeffrey MacDonald, 26. He told them he’d been awakened to the screams of his family, and was immediately set upon by scruffy intruders who were chanting, ”Acid is groovy – kill the pigs.” He lost consciousness, he said, and when he awoke the intruders were gone and his family was dead. ”Pig” was written in blood on the headboard of a bed.

The doctor’s wounds were trivial, at least compared with those of his family. He had a small, neat incision between his seventh and eighth ribs, just deep enough to partially collapse a lung. He had a lump on his forehead, a cut on his left arm and some superficial lacerations. Only the lung required treatment.

Almost from the start, investigators focused on MacDonald, whose account of the attack seemed to contradict the physical evidence: the locations of bloodstains and spatters and of torn fibers from his pajamas, and the lack of evidence of a furious defensive struggle. Although the doctor was initially cleared by a preliminary Army hearing – ”insufficient evidence” – he was prosecuted in 1979 in federal court, where he was convicted after a six-week trial. The jury was out only six hours. The prosecution’s methodical use of circumstantial evidence had overcome a signal weakness in the case: The state never presented an entirely convincing motive.

The lurid crime was an international story that became even bigger with the publication in 1983 of ”Fatal Vision,” the bestseller about the case by journalist Joe McGinniss; the book would lead to a TV miniseries and, later, to an excruciating spasm of self-examination by the media. The MacDonald murders had spawned a second Big Story.

From prison, MacDonald sued McGinniss for breach of contract, alleging that the writer had betrayed him. During trial and its preparations, McGinniss had been embedded with the defense team, ostensibly to write a book about an unjustly accused man. At some point, the writer became convinced of the doctor’s guilt but never told MacDonald, fearing he would lose the doctor’s cooperation. So, as he was ingratiating himself with the convicted killer, feigning incredulity over the verdict, encouraging MacDonald to send deeply self-revealing letters from prison, McGinniss was writing a book describing MacDonald as a narcissistic, psychopathic monster. McGinniss contends he behaved ethically in dealing with MacDonald under enormously difficult circumstances.

The civil jury deadlocked – McGinniss’ publisher wound up settling out of court – but after reviewing the trial transcripts and exhibits, New Yorker writer Janet Malcolm in 1990 wrote an article that eviscerated both McGinniss and the entire profession of journalism. Its famous first line: ”Every journalist who is not too stupid or too full of himself to notice what is going on knows that what he does is morally indefensible.”

Malcolm contended that McGinniss’ tactics were symptomatic of what all journalists do, to some degree: fool people into trusting them, then betray them by spinning facts, or distorting them, to create whatever compelling narrative they wish. Every story, she implied, is on some level a con job.

The accusation rattled the prickly, inward-peering world of journalism. Reporters took sides – most in a defensive crouch but some acknowledging an unnerving whiff of truth in what Malcolm wrote. Journalists do have agendas; their articles do marshal facts selectively. They don’t always reveal their biases. Stories are seldom completely arm’s-length. This greater journalistic debate seemed inextricably linked to the MacDonald case; it became almost impossible to write about one without summoning the ghosts of the other.

So, just to be clear:

I intend to spin you toward a certain conclusion. The process is stealthy and has already begun; it was no accident that I called Dr. MacDonald’s stab wound an ”incision.”

I have come to believe that Jeffrey MacDonald murdered his family and injured himself as part of a coverup; I’ve concluded this both because I have researched the case extensively, and because, as a writer, I see exactly how Errol Morris prejudiced his account while shrewdly appearing not to do so. I admire his skill but not his book. I think the media have been careless and gullible in reviewing it, perhaps partially because the story of a grievous, enduring miscarriage of justice presents a more compelling narrative than the alternative.

I think ”Fatal Vision” is among the best true-crime books ever written, but I believe Joe McGinniss unattractively betrayed Jeffrey MacDonald to keep the doctor talking. Still, I do not think McGinniss deserved the national scorn he endured. The partnership between the journalist and the murderer was an exercise in ferocious, mutual exploitation, for enormous stakes, and MacDonald’s lie – that he was innocent – was a far greater deception. And, anyway, in such a freighted transaction no meaningful measure of morality attaches.

I’d never met Brian Murtagh before starting on this article, but he and my lawyer-wife worked together for years at the Justice Department. She would tell me about her friend who, 30 years after the case that would consume his life for better or worse, could not stop talking about ice picks, bloodstains, holes in pajamas, and beautiful, dead children.

OK, I think we’re at arm’s length now. We’re good to go.

Helena Stoeckley was a mess. In 1970, the Fayetteville, N.C., townie was 19, a bloated, blowsy drug addict who ran with a shabby crowd and whose daily consumption of narcotics, booze and hallucinogens – by her own accounting – bordered on the suicidal. Well over half of Errol Morris’ book argues from one angle or another that Stoeckley was probably in the MacDonald apartment that night – that she was the intruder described by MacDonald as a woman in a floppy hat who was chanting and holding a candle. To believe MacDonald innocent by the central thesis of the book, you must believe Stoeckley guilty. It’s certainly tempting. She confessed early and often.

In the days after the murders, Stoeckley told friends that she might have been present at the crime. She’d been out with others that night, she said, and though she was so smashed on LSD and mescaline and had no clear memory of where she had been, the details of the crime, and MacDonald’s story of hippie intruders, sounded familiar to her. She remembered maybe holding a candle dripping blood. She had vivid dreams of the crime scene. A polygraph examiner reported that ”in her mind” she’d been there.

But when Helena Stoeckley was located, detained and seated on the witness stand at the 1979 trial, she flatly denied any knowledge of the crime. It was a disaster for the defense.

In the years afterward, as interest in the case continued and she’d attained a degree of perverse local fame, Stoeckley decided she’d definitely been in the house. To private detectives hired by the defense, she even named accomplices. She supplied motives.

There were, however, serious discrepancies in her varying tales. One person she’d fingered as Colette MacDonald’s killer happened to have been in prison on the day of the crime; Murtagh produced the jail records. Stoeckley told private investigators that she and her friends had come to the house to hassle MacDonald for drugs, and that when the doctor refused, the violence ensued. This claim is duly reported in ”A Wilderness of Error,” but Morris doesn’t step back to consider the singular, contravening fact that Jeffrey MacDonald himself had never claimed he’d talked to the supposed intruders at all.

Morris finds it deeply suspicious that police never took Stoeckley seriously. It might have been more surprising if they had. Over time, her claims got more and more outlandish. She told one man that she now remembered she’d known both Jeffrey and Colette, had applied for a babysitting job at the house, and had once gotten high with the doctor and had sex with him.

(Over the years, the government would test crime-scene evidence against Stoeckley’s fingerprints and DNA, and the fingerprints and/or DNA of several of the men she’d most frequently named as accomplices. No matches.)

This fall’s hearing in Wilmington was held partly to consider an allegation involving Stoeckley, who died of cirrhosis in 1983. Seven years ago, a former assistant federal marshal named Jimmy Britt – also now dead – signed an affidavit for the defense saying that Helena Stoeckley unambiguously confessed to him that she had witnessed the crime, an admission made, he said, during an hours-long drive when he brought her from South Carolina to testify at the trial; the next day, he said, he saw her admit this same information to a prosecutor (not Murtagh). Britt said the prosecutor told Stoeckley that if she testified to being at the MacDonald house that he’d charge her with murder. If true, that could be considered prosecutorial misconduct, resulting in a new trial. Britt said he’d finally come forward because of a guilty conscience.

The story of the Britt affidavit was in ”A Wilderness of Error” to explain why Stoeckley might have lied on the witness stand in the original trial. But in Wilmington, the allegation was convincingly refuted when prosecutors called their rebuttal witnesses.

None of what was about to come out was in Errol Morris’ book, though it was available to him in public records.

Jimmy Britt, evidence suggested, had not transported Helena Stoeckley from South Carolina at all; he’d never had hours to talk to her. The transport had been done by a tag-team succession of other marshals. Some of the paperwork still survived. Two of the transporters testified.

Had Britt lied about the whole thing? His old bosses testified he was a chronic malcontent who sought the limelight; the implication was that he might be settling old scores with a made-up disclosure embarrassing to his supervisors. But how could he have thought he’d get away with it?

Murtagh proposed an answer that was characteristically precise, nerdy, complicated and effective. He entered into evidence a document he had unearthed. It seemingly had nothing to do with the case – it was just a routine receipt mailed to the U.S. Marshals Service in 2002 for records the service had sent to be archived at the Federal Records Center. The receipt, though, contained a notation that Marshals Service records now would be destroyed after 25 years, instead of the previously standard 55. The recipient of the letter was a Marshals Service administrator who happened to be Jimmy Britt’s wife, Murtagh showed.

What did this all mean? It meant, Murtagh explains to me later, that in 2002, Jimmy Britt might have had reason to believe that by 2005, records relating to the 1979 transport of Helena Stoeckley would have been destroyed, making it almost impossible to prove one way or another which marshal had done it. After remaining silent about this for a quarter-century, Jimmy Britt waited until exactly 2005 to come forward with his affidavit. (That records survived apparently was an administrative oversight.)

Jeffrey MacDonald hates Brian Murtagh.

Outside the courthouse, Colette’s brother, Bob Stevenson, is exulting, declaring the Britt affidavit ”crushed, completely demolished!” Kathryn MacDonald tells me she’s sure Britt had been telling the truth, and predicts it will become obvious to everyone when the defense answers the prosecution. She adds that the judge in the case, James C. Fox, won’t be fooled by this smoke screen, since he trusted Jimmy Britt, once even adopting a puppy from him.

But the defense never answered the prosecution on the Britt affidavit. I asked the judge’s assistant about Britt and a puppy. She checked with the judge: no puppy.

Murtagh today is unrecognizable from the news photos of him on the day of the verdict 33 years ago – a skinny nebbish in front of microphones, dressed in the unflattering florid polyester of the time. It was only his second trial. (He’d lost the first, against a distributor of dirty magazines.) During the final tense mornings of the trial, Murtagh would wake up, then throw up.

Responsibility for the huge case had fallen into his inexperienced hands by default in 1975, when the original lead prosecutor, Victor Worheide, died of a heart attack.

Murtagh has had an unconventional career: Half his professional life, he estimates, has been consumed by two cases only: the prosecution of Jeffrey MacDonald and the prosecution of the Libyan intelligence officer who masterminded the explosion of Pan Am Flight 103 over Lockerbie, Scotland. He won them both.

Murtagh and I are in his living room. He’s remembering how he first got involved with the MacDonald case, back in 1971, as an Army lawyer. A police investigator was showing him the crime-scene photos, trying to urge him to join the prosecution. They got to a certain picture, the one he and I are looking at right now.

It is of Kristy, the 2-year-old. She lies curled up in her bed. Beside her, on the floor, is a tawny-orange, googly-eyed stuffed toy dog, bigger than she is, turned to face her as if sadly bearing witness. There is blood puddled at her abdomen, dripping onto the floor.

”There were wounds in Kristen’s chest that didn’t have corresponding defects in the clothes,” Murtagh says. He talks in lawyer speak. What he means is: Kristy was probably killed as she slept, by someone who first lifted her pajama top, as if to better identify the location of vital organs. She is knifed front and back, with anatomical precision, along the heart, the aortic arch and the pulmonary vein.

Back in 1971, the investigator looked at Murtagh looking at this picture, and said, ”You’re in.” It wasn’t a question; it was an observation.

The strongest case for Jeffrey MacDonald’s innocence? From the very beginning, his story never varied. The strongest case for Jeffrey MacDonald’s guilt? From the very beginning, his story never varied. It locked him into a specific narrative that would prove problematic.

MacDonald said Kristy had wet his side of the bed, so he’d gone to sleep on the living-room couch, where he was assaulted. His pajama top had been pulled up over his head and was stretched between his forearms, he said; he tried to use it to fend off the attack, but was clubbed and stabbed into unconsciousness. When he came to, he was alone with the corpses of his family.

Because of a strange coincidence – each member of MacDonald’s family had a different blood type – police could easily map who had been attacked, and where. The tale of the blood was very different from the tale MacDonald told. There were many significant discrepancies, among them that the doctor’s pajama top was stained with Colette’s blood before the garment was torn. That should have been impossible if, as MacDonald said, it was torn in the initial struggle with the intruders.

”A Wilderness of Error” doesn’t dwell long on the blood evidence. Citing shoddy detective work – much of the initial police work was, indeed, sloppy – Errol Morris basically dismisses all of it as tainted. Outside the courthouse in Wilmington, he tells a camera crew that he doesn’t see ”a shred of evidence” suggesting MacDonald’s guilt.

It was an odd choice of language, considering. Shreds of MacDonald’s torn pajama top were central to the case against him: Broken pajama threads were not found in places they should have been if his story were true – near the sofa on which he was allegedly attacked, for example – but were found in places they should not have been, such as beneath Colette’s body.

And then there was Murtagh’s smoking gun – his big triumph at trial.

McDonald had told police that after he woke up, he put his pajama top on Colette’s chest. Forensics showed that 21 ice pick holes in Colette’s body lined up perfectly with 48 holes in Jeffrey MacDonald’s pajama top, when the pajama top was folded a certain way so that some punctures went through more than one layer of fabric. Murtagh argued that the only explanation for this was that MacDonald had delivered those blows himself to a dead or dying Colette. The jury bought it. But ”A Wilderness of Error” finds this preposterous – the prosecution never established that this was the only way the pajama could have been folded to get that pattern of holes, Morris argues; conceivably, you could manipulate any piece of cloth into innumerable shapes that would produce the same pattern.

I caught up with Morris during a break in the hearing, and told him that I thought he was wrong: I described an experiment I had done, in which I used two sheets of paper and a dart. I folded the upper sheet on itself, irregularly, like the pajama top, and then stabbed through it, counting carefully so as to make exactly 21 holes in the lower sheet but 48 in the upper.

Then I tried to fold the upper piece of paper in a different way, so that all the holes still lined up. It becomes obvious almost immediately that you just can’t do it. Each change you make radically alters the relationship of one puncture hole to all the others.

Morris listened, nodded.

”That remains on my mind,” he said.

Morris said he’d suspected that might be true and worried about it a little, but then decided it didn’t matter, because even if it seems ”nonsensical” that the holes in Jeffrey’s pajama top align with the holes in Colette’s body through mere coincidence, the alternative is equally nonsensical: That Jeffrey MacDonald, while trying to get away with a murder, would be stupid enough to stab his wife through his own garment. Think about that, he urged me.

I promised him I would.

I have.

Here’s what I think: I think it is not all that nonsensical to imagine that in the immediate, terrifying aftermath of having made the most unwise decision in his life, MacDonald might have made another.

There are many significant, incriminating facts glossed over in, or completely omitted from, ”A Wilderness of Error.” Conversely, much is made of nonsense. An entire chapter is devoted to the supposedly startling fact that Helena Stoeckley reported seeing a broken rocking horse in Kristy’s room. Yes, the horse had been clearly visible in newspaper photos, but no one, Morris argues, had ever publicly disclosed it was broken.

There may be a good reason for that: It doesn’t appear to have been broken. Murtagh and his colleagues demonstrated that effectively in Wilmington, producing a rocking horse of identical design and comparing it with crime-scene photos of the original. There’s only one way to disable that horse, and it’s by disconnecting a spring. If you do, it lists dramatically to one side; the toy at the crime scene was upright. In Morris’ book, the only confirmation of the claim that the toy was broken was from a close friend of Jeffrey MacDonald’s mother.

Without context or comment, the book quotes a man recalling that a doctor once said the stab to MacDonald’s chest came so infinitesimally close to his heart that even a cardiac surgeon wouldn’t have risked self-inflicting it. Unlikely: McDonald’s stab wound was to the right side of his chest, as photos clearly show, a fist’s breadth away from his heart.

Morris mentions that DNA tests on a hair found under Kristy’s fingernail showed no known source, meaning it could have been from an intruder. This was the second evidentiary claim made in the Wilmington hearing. Murtagh presented affidavits from lab technicians showing it was a naturally shed hair, not the result of violence; also, that it had not been found on Kristy at all, but had resulted from accidental contamination of the lab sample. The defense didn’t effectively refute this.

But there’s another, more important, hair sample that Morris’ book doesn’t spend much time on at all. He draws no inferences from it. It is a broken hair from a male Caucasian, with blood on it, found at the crime scene in Colette MacDonald’s hand. At trial in 1979, experts could not positively identify its source, and defense lawyers made much of this failure, suggesting this was proof of an outside murderer. The imprint of the Real Killer!

This was one of the hairs analyzed many years later at the behest of the defense, under more sophisticated tests, with clearer results.

The hair was Jeffrey MacDonald’s.

Many of the supposed disclosures in ”A Wilderness of Error” are not new – the original jury heard them, weighed them against the prosecution’s competing evidence, and sent MacDonald away for life. Several of the trial jurors are still around, but Errol Morris didn’t talk to any of them.

”He got a fair trial,” Fred Thornhill says, ”and he got what he deserved.”

Thornhill is 60 now, still a general contractor in Raleigh, N.C. At the time of the trial he was 27, and he very much wanted to acquit Jeffrey MacDonald.

”We all did,” he said. The alternative, he said, was to imagine a father capable of such a crime. No one wanted to go there.

”We heard the prosecution’s case, which was pretty credible,” Thornhill remembers, ”and then we waited for the defense to blow it out of the water. They never came close. They couldn’t refute the physical evidence.”

Thornhill found Helena Stoeckley not remotely credible – ”a lost person, really pitiful” – and says he likely would not have believed her even if she’d said she’d watched the murders.

A major problem for Jeffrey MacDonald, Thornhill said, was Jeffrey MacDonald:

”He was a very egotistical person, and it absolutely came through. When he took the stand and started that phony crying, we were astonished. It was like bad acting. When we got to the jury room, we weren’t allowed to discuss it, and we didn’t, but we’re looking at each other, and it was like, ‘What just happened out there?’

”You know, there was testimony that a month after [the murders], MacDonald had sex with a nurse. The defense objected, and it was stricken from the testimony, but how do you strike that from your mind?”

That was the thing, Thornhill said. The jurors felt it: There was something fundamentally wrong with Dr. Jeffrey MacDonald.

Here’s a riddle: At the funeral of her sister, a woman meets a man and falls in love with him. But she never asks his name and loses track of him, and when the funeral is over, he is gone. No one can identify him.

Two weeks later, the woman murders her brother. Why?

All essential facts are known to you. Any guess?

The answer: Because she thinks the man might come back for the brother’s funeral.

This is said to be a primitive psychological test to detect sociopathy. A sociopath, who is amoral and makes decisions solely based on his own needs, supposedly would see the answer immediately.

Would Jeffrey MacDonald get it right? Joe McGinniss thinks so. ”Fatal Vision” finds ample evidence in MacDonald’s behavior alone.

Not long after the murders, MacDonald called his father-in-law to say he’d tracked down and killed one of the murderers. It was a lie, as he later admitted, to get his father-in-law off his back. He’d wanted to move on, he said, but Colette’s father wanted him to pursue justice.

Examples of MacDonald’s creepiness drench the pages of ”Fatal Vision,” but they are almost completely absent from ”A Wilderness of Error.”

After initially being cleared by the Army, ”Fatal Vision” says, MacDonald lived for years as a civilian doctor in California, where he led an uninhibited life: a yacht, fast cars, many sexual partners. If this was the life he craved, it was one he never would have had as an Army doctor with two children and one on the way, and a wife on whom he was serially cheating and whom he had married because she was pregnant.

The night of violence probably began in a rage, but at some point, by the prosecution’s theory, the killing became methodical. I asked McGinniss why he thought Jeffrey MacDonald was able to do this to his family:

”Because,” he said, ”they’d become impediments.”

The doctor, perhaps, wanted a do-over.

Unlike Murtagh, McGinniss has found a way to break free from this case. He has written other books, most notably ”The Rogue,” a 2011 tell-all account of the life of Sarah Palin. Palin hated it, and when Morris’ book came out, Palin wrote a glowing online review of it, taking no position on MacDonald’s guilt but opining that McGinniss is exactly the sort of conscienceless scoundrel to hound an innocent man into prison for life.

So, maybe Joe hasn’t entirely broken free from this case, either.

It’s definitely not out of his mind. He told me something I had not known: Not long after the murders, Jeffrey MacDonald asked police for permission to go back into his house, unaccompanied. It had been sealed as a crime scene. The request was denied.

Why do you think he wanted to do that, I asked.

”They never found the scalpel blade he used on himself,” McGinniss said. ”I think he knew where he’d hidden it. Maybe between floorboards. I think he wanted to get rid of it.”

”He hadn’t slept for 24 hours, and he was taking amphetamines to lose weight. It’s a bad combination.”

We are in Murtagh’s living room, surrounded by case files. I’ve asked him to go through the night of the murders, based on the prosecution’s case – blood, fibers, testimony, informed conjecture.

When the doctor went to bed, Murtagh said, Kimberly was there, and she had wet his side of the bed. MacDonald, amped on speed and exhaustion, exploded.

I interrupted. Wasn’t the bed-wetter Kristen, the 2-year-old?

”That’s what MacDonald claimed, but the urine turned out to be consistent with Kimmy’s blood type.”

Why would he lie about that?

Two reasons, Murtagh said. First, it would better fit his story of who was where ”when the hippies fanned out across the house. Also, because it would seem less plausible that a fight would break out with his wife over a 2-year-old wetting the bed than a 5-year-old.”

He was thinking that analytically?

”By the end, he was. When he was moving bodies around.

”So, Colette and Jeffrey have a confrontation over the bed-wetting. We know it was an issue between them because Colette had mentioned it in a child psychology class she was taking. He’s whacking Kimberly, and Colette tries to defend her.

”He goes and gets the club. Colette has a pattern bruise with the end of it on her breastbone. So, he’s using it like a bayonet thrust to keep her away. There is something she’s holding that he doesn’t want to get in contact with. It’s the knife with the bent blade. I think she got it from the dresser. Its blade is consistent with a cut on his left sleeve, but on no other injury to anyone.

”Then, he swings the club. I’m willing to give him the benefit of the doubt. I don’t think he was going for the girl. I think he swung at Colette but he missed and cracked Kimmy’s head open like a walnut. She falls and bleeds copiously on the shag rug.

”Colette screams, ‘Jeff, why are you doing this?’ It’s loud enough that he thinks a neighbor might have heard it, so he’ll include it in in his story, but he turns it into, ‘Jeff, why are they doing this?’ The true version makes a lot more sense.

”At this point, he has to render Colette unconscious, because she has just seen her child mortally injured, and she’s going to fight to the death. This is where she grabs Jeffrey and tears his pajama top. This is where both her arms are broken, with massive blunt trauma to the head and face. She’s down.

”Now, he’s really in it and has to figure out how to get out of it. He goes into the living room and breaks down. An upstairs neighbor hears what she identified as either laughing or crying. There’s a magazine on the coffee table with a story about the Manson murders. He gets his idea.

”What I think happens next is that Colette gets up. She’s staggering into Kristen’s room to try to get her out of the house. Jeffrey sees, follows her in and clubs her from above. Her blood spatter is on the wall, consistent with centrifugal force from the club. He’s knocked her out again. Kristen is probably still asleep.

”Now, you have to start moving bodies to make it seem as though everyone was attacked in her own bed. You transport Colette in a pile of bedding, get her blood on it, from an imprint of your own sleeve.”

I couldn’t help notice that Murtagh had slipped into the second person.

”Now, you’re improvising. You must support the story of multiple intruders. . . . ”

Murtagh won’t really admit this is personal to him, that it’s really under his skin. That would be un-lawyerly. Ask him why he has stuck with this case so long, pressed so vigorously to keep Jeffrey MacDonald in prison, and he quotes federal case law about the appropriate penalty for first-degree murder.

But it is personal. This case made Murtagh famous, but it also, ironically, circumscribed his career. Despite his formidable talents, to stay with this case, he has had to remain a federal prosecutor, at government salary, his whole life. He has been the institutional memory of the case, indispensable to fighting a relentless series of appeals, one of which was deeply personal, charging him with misconduct for supposedly suppressing evidence. He won that, too. He has stayed on because of an abstract sense of justice, but also because of a concrete duty he feels to three people who died very badly.

”So,” he continues, ”you get the Old Hickory knife from the kitchen, and the ice pick, because now you’ve got to make this convincing. This has to seem like a frenzy. And you go to work on them.

”You probably do Kristy last. She is still asleep. The first thing she knows is when you go through her chest.

”You put on a surgeon’s rubber glove. You write ‘Pig’ in blood on the headboard of the master bedroom, without leaving fingerprints, but you accidentally shed a thread from your torn pajama top near it, where it shouldn’t be.

”Then, you go into the bathroom. You get a scalpel blade, and you do yourself in front of a mirror. You’ll drip your own blood there.

”Then, you call the police.”

I had one question: If Kristy slept through it all, why did he have to kill her?

For the first and only time, Murtagh’s composure slips a little. He starts to speak, stops, swallows and starts again.

”He had to do it,” Murtagh said, ”because it’s so unthinkable. That was the one, the cold one, that no jury would believe a father could do.”

Jeffrey MacDonald and his lawyer declined to answer questions or comment for this article.

A ruling on the claims made in the evidentiary hearing in Wilmington is expected early next year. If the ruling goes against him, it will probably be appealed to a higher court. And if that court denies the appeal, Jeffrey MacDonald might be out of options, finally.

Just before this story went to press, Errol Morris and I spoke for nearly an hour; he concedes there are some things he wishes he’d written differently – for example, disclosing that there were some credible challenges to Jimmy Britt’s story. Morris allows that he may have used some facts selectively to make a case for what he believes – selectivity, he says, is part of all journalism – but adds that his belief remains solid that MacDonald did not get a fair trial. He also thinks MacDonald is innocent, but of that is less certain.

I asked Brian Murtagh if he has any regrets about sticking with this case, and he said no. I asked him if he has any nightmares about it, and he said he used to dream about the crime scene, but no more. Only one dream still visits him from time to time.

”I find myself in an emergency room, badly injured, and I look up at the doctor about to work on me, and it is Jeffrey MacDonald.”