Scott Macklem was twenty years old. His murder has never been solved.

The paper trail begins at the hospital on the morning of the murder, and they contradict each other. Officer Jim Carmody's report from that day says the victim's fiancée's sixteen-year-old sister came to him unprompted with the name of a man she knew as John Lamar, her sister's former boyfriend, whom the family believed responsible. Sergeant John Bowns's report, written the same day about the same emergency room, credits the victim's boss instead, saying Scott had told him of a threat weeks earlier, and says nothing about the sister. Years later the boss recalled it differently, remembering that the fiancée gave the police the name. The fiancée's own account is different again: no one in her family volunteered it, an officer asked her whether she had former boyfriends, and she answered. Four accounts of the case's first hour, and no two of them match.

Before Freeman was a suspect, he pointed police to another man and a plate; no one ever showed him that man's photograph.

Later on the day of the murder, the identification witness told an officer that a man he saw in the parking lot resembled the driver. He testified it was "more or less a possibility," seen from too far away "to make a positive identification." The plate he gave traced to a Florida car belonging to James Kilbourne. Det. Hudson agreed at trial that the witness had identified Kilbourne that day. Sgt. Bowns testified the witness "said he saw a man that returned to the parking lot that looked similar," and that he never showed the witness a photograph of Kilbourne: "I didn't have a picture of Mr. Kilbourne."

Nothing in the record connects James Kilbourne to this murder, and this page makes no such claim. The exchange bears on the reliability of the identification the jury was given.

No one saw the shooting. A student in the lot heard the blast and saw a car pass him at ordinary speed on its way out. He got a look at the driver, which he estimated at about five seconds, and about ten minutes later wrote down a plate number. It came back to a Buick sold eighteen months earlier, and he told the jury it came back wrong.

What that witness described, and what a hypnosis session did to it that same morning, has its own card below.

Start with what hypnosis does to a memory. A hypnotized witness tends to grow more confident and more suggestible at the same time, so a detail introduced by the person asking the questions can come back as if it had been remembered. That is why courts treat hypnotically assisted identification with caution, and it is what the session transcript in this case shows happening.

The witness was hypnotized on the morning of the murder by Dr. Thomas Mooney, a psychology instructor at the college who, by his own present-day account, had studied hypnosis but used it on almost no one. He recorded the session; the recording is missing, and only the transcript survives. Under hypnosis the witness described the car as “light tan in color, apparently mink color,” a foreign hatchback, and when the questioning turned to its make he was asked, in substance, whether it was an American car and answered no. The driver he described was wearing a navy ski mask worn up like a stocking cap; he said nothing about facial hair until the hypnotist asked, with no prompting, “Now he’s got a beard too?” He also described how the victim looked as he fell, though he had already said he did not see him fall.

The number he later wrote for the license plate was assembled the same way, through a run of proposed letter combinations, and it traced to a car that had nothing to do with the case. An expert who reviewed the session transcript years later called it close to the worst hypnosis session he had seen.

What the session produced, the trial then hardened. The witness told the jury, “I know it was him driving the car,” while conceding the plate “came back the wrong plate number, I know that.” Certain of the conclusion; uncertain, by his own account, of the car, the plate, and the face.

A shotgun shell box recovered in the investigation bore a latent fingerprint. The state’s own 1986 laboratory report excluded Freeman as its source.

The box was a Federal 20-gauge shell box; the fatal round was 12-gauge (Tr. 955). The difference is a reason to test, not a reason not to.

In April 2008, at Port Huron PD’s request, the State Police searched the print through Michigan’s AFIS and the FBI’s IAFIS: no subject was developed, and the report closed with one sentence that matters now: “The latent print will remain on file.”

That was eighteen years ago. The FBI’s current system, NGI, did not exist until 2014, its latent-matching technology is a generation better, and the databases have grown by tens of millions of prints since the last search. The print is sitting in a state file with a laboratory number, waiting to be searched again.

The failures began at the scene. By a witness’s account, no crime-scene tape went up for at least three quarters of an hour, on a working campus parking lot with people moving through it. The department photographed the scene roughly fifteen times. Nothing the victim was carrying was tested: not the gym bag beside his car, not the textbooks on the hood of it, not his clothing, and the jacket the first person to reach him pressed to the wound was never collected. A police report names seven young men standing in that lot the morning of the murder; the record shows no interview of any of them. The man who reached Scott Macklem first, and tried to keep him alive, got a single follow-up call a week later and was never called to testify.

Two more absences sit in the file itself. No toxicology was performed. And no interview of the victim’s brother appears in the investigative file.

Almost every police report in this case was written and submitted by one man: Sergeant John Bowns, for whom Scott Macklem’s murder was a first homicide as lead investigator. Four years before the murder, the Port Huron Police Department had terminated Bowns for misconduct and neglect of duty after the State Police caught him gambling illegally; he was later reinstated.

Herb Welser, the retired Port Huron detective lieutenant who has re-investigated this case, set out to trace Bowns’s paperwork and reached an unsettling conclusion: a number of the reports cannot be corroborated. People named in them as having been interviewed have said the interviews never happened. The motive theory that Freeman learned of the engagement at a Halloween party rests on Bowns’s reports; the people those reports place at the party do not recall speaking to police. Another detective on the case, Harry Hudson, reviewed a report bearing his own name that described interviews he and Bowns supposedly conducted together, and said they never took place.

Bowns died two decades ago. The point is not his character; it is that the documentary spine of this case is the work of a single investigator whose reports, examined four decades later, do not all hold up.

Two weeks after the murder, the Michigan State Police post near Escanaba reopened its assistance file for one stated reason: the suspect “had alibi witnesses in the Escanaba area, to the fact that he had been in Escanaba on November 5, 1986.” A trooper began interviewing and listed further witnesses to contact.

The journal entries that follow tell the rest: a call to the Port Huron detective goes unreturned; then, February 11, 1987: “Reviewed, close this out.” Final disposition: “our help was not requested or needed.” The trial began ten weeks later.