John Norman Collins was convicted in 1970 on forensic methods that would not survive peer review today. The judge admitted them. The prosecutor built his case on them. Six other families never got a trial at all. Thirty-five years later, the same state laboratory that produced the original forensic record contaminated the DNA evidence that was supposed to reopen the case. The Michigan Murders were never closed. The record just stopped asking questions.
Seven Murders. One Trial. Zero Answers.
Between July 1967 and July 1969, seven young women were murdered in and around Washtenaw County, Michigan. The victims were students, young women who hitchhiked or sought rides home, women whose names the public eventually stopped saying out loud. The task force assembled to find the killer was a jurisdictional mess: the Ann Arbor Police Department, Ypsilanti Police, Washtenaw County Sheriff’s Office, Michigan State Police, Eastern Michigan University Campus Police, and University of Michigan Police operating without a shared database, a unified command structure, or mandatory cross-agency data sharing.
What that produced was not just disorganization. It was information entropy. Tips were not simply lost in the shuffle. They were actively withheld as agencies competed for the arrest. The “collar” carried professional credit. Sharing information meant sharing that credit. The result was a task force in name only, six agencies pointed at the same problem and pulling in six different directions. This is the exact same structural pathology that would cripple the Oakland County Child Killer investigation seven years later. The lesson was not learned because it was never acknowledged. The system filed it under “communications challenges” and moved on.
John Norman Collins, a senior at Eastern Michigan University, was arrested in July 1969 and charged with the murder of Karen Sue Beineman. He was tried in 1970 before Washtenaw County Circuit Court Judge John W. Conlin. Washtenaw County Prosecutor William F. Delhey led the prosecution. The jury deliberated for 27 hours across four days before returning a first-degree murder verdict. Conlin sentenced Collins to life in prison without possibility of parole.
That was the only trial. Six families got nothing. Delhey was explicit about the calculus: the office pursued the case with the strongest available evidence. What he did not say, and what the record makes plain, is that the strongest available evidence was not particularly strong. It was the strongest evidence produced by a forensic framework that was already beginning to crack.
What the Science Actually Was
The prosecution’s forensic case rested on two methods. The first was Neutron Activation Analysis, which examines the elemental composition of hair samples to establish similarity. A graduate chemist from the Washtenaw County Health Department testified that hair found in the basement of Collins’ uncle, MSP Corporal David Leik, was chemically consistent with hair found on Beineman’s clothing. The second was microscopic hair comparison, which matches physical characteristics of hair strands under a microscope.
Judge Conlin admitted both. Defense counsel Neil Fink objected. The jury heard the testimony as scientific proof of connection. Collins was convicted.
That sequence deserves a closer look. Once a judge puts their imprimatur on a forensic method, the jury stops examining the logical gaps in the evidence and starts looking for permission to decide. The NAA testimony gave them that permission. It was presented as if it were a digital fingerprint. It was not even close to a digital fingerprint. But Conlin’s admissibility ruling transformed a contested scientific technique into a trial-grade fact, and the jury processed it as such. The defense objections were on the record. They did not change what the jury heard.
The FBI abandoned Neutron Activation Analysis for hair identification in the 1970s, shortly after the Collins conviction, because the elemental composition of hair varies too widely within a single individual to function as an identifier. A 2015 Department of Justice review of FBI forensic practices found that examiners had provided scientifically unsupportable testimony in approximately 90 percent of cases involving microscopic hair comparison. The DOJ review covered 268 trials. In 257 of them, the hair testimony was flawed.
Conlin’s ruling admitting the NAA testimony was the evidentiary foundation of the conviction. Not the only evidence, but the scientific anchor. Without it, the case against Collins was circumstantial proximity. With it, the prosecution had something that sounded like certainty in a courtroom where certainty was the only thing that was going to calm a terrorized county.
The transcripts of Conlin’s admissibility rulings have since been purged from Washtenaw County court records. Researcher Gregory Fournier documented this through FOIA requests. What survives is approximately 800 pages of news clippings from 1967 to 1970, archived testimony summaries, and Ann Arbor News reporting from the trial period. The purging of the transcripts means the precise legal reasoning behind the admissibility decision is no longer recoverable from primary sources. What is recoverable is the result: the jury heard junk science presented as forensic certainty, and it convicted.
Delhey’s Choice and What It Cost
William F. Delhey served as Washtenaw County Prosecutor for 29 years. He tried the Collins case himself, alongside assistant prosecutor Booker Williams. He made the decision not to prosecute Collins for the other six murders, and he made it deliberately. Twenty years after the trial, Delhey told a reporter he remembered everything about the cases without consulting files. He said of the decision: “We took the case we had the best chance of winning.”
That is an honest answer. It is also a damning one. The Beineman prosecution was the most expensive criminal proceeding in Washtenaw County history to that point. Delhey acknowledged both the cost and the strategic rationale. The subtext is that six other prosecutions, each requiring the same evidentiary infrastructure against a defendant with capable defense counsel, were not winnable on available evidence.
Delhey’s decision to limit prosecution to the Beineman case produced a structural outcome that served the office’s interests without serving all of the victims’ families. Collins received a life sentence. The prosecutor’s office recorded a conviction in the most high-profile criminal case in Washtenaw County history. Six cases were effectively closed without being tried. The public felt safe. The families of Mary Fleszar, Joan Schell, Jane Mixer, Maralynn Skelton, Dawn Basom, and Alice Kalom did not get to watch their daughters’ presumed killer answer for those murders in court. They never will.
Delhey died in October 2000 after 29 years in office. He never faced discipline. The Prosecuting Attorneys Association of Michigan named an award in his honor, given annually for integrity and lengthy service. The families of six victims are not among the award’s documented beneficiaries.
The Keyes Problem
Edward Keyes was a New York journalist who wrote “The Michigan Murders,” published in 1976. It remains the definitive narrative account of the Collins case. It is also, in retrospect, a lesson in what happens when a skilled journalist treats law enforcement’s synthesis as the record.
Keyes was not lazy. He was thorough, sympathetic to investigators, and careful with the emotional stakes of the story. He used pseudonyms for victims and the convicted killer in an era before widespread internet access, reportedly to shield families. What he could not do, writing in 1976, was subject the prosecution’s forensic claims to independent scientific review. He had no independent access to the lab data. He had the trial record and the investigators’ accounts. The investigators believed Collins killed all seven women. Keyes wrote as though they were right.
The problem is that Keyes noted the anomalies himself. He documented that the Mixer case differed from the others in significant ways: Mixer was not sexually assaulted, she had not been hitchhiking, her body was found quickly and was not subjected to the same post-mortem treatment as the others. These were not minor variations. They were the kind of signature-level differences that should have prompted independent analysis of whether the same perpetrator was responsible. Keyes flagged them and kept going. He folded Mixer into the Collins narrative because the investigators said she belonged there.
That is where the structural problem lives. The prosecution and the media had created a gravitational field around Collins so powerful that anomalous data points were pulled into orbit around him rather than treated as independent signals. When the evidence did not fit, the explanation was variation in method, not a different perpetrator. It was the ultimate failure of the one-size-fits-all monster narrative: focusing so hard on the killer you believe you have caught that you stop reading the data points that do not fit. Whoever actually killed Jane Mixer operated in that shadow for decades, while the Collins hysteria held the frame closed.
Keyes was not a bad journalist. He was a journalist operating in 1976 without DNA technology, without independent forensic review, and without the institutional skepticism that comes from watching prosecutors build careers on questionable science and then watch those careers survive intact. He trusted the record that law enforcement handed him. That trust produced a book that treated a contested evidentiary picture as a settled narrative. When the 2005 DNA results proved Keyes’ own instinct was correct and his conclusion was wrong, he was not alive to correct the record. He died in 2002. The book was reprinted in 2010 with an updated epilogue noting the Mixer exclusion.
True crime journalism’s relationship with prosecution is a longer conversation, and this series will have it. The Keyes problem is not unique to Keyes. It is structural. A writer working from a closed case, interviewing investigators who believe they got the right person, has a built-in evidentiary ceiling. The investigators control the record. The record shapes the narrative. The narrative becomes the public understanding of the case. When the science breaks, the narrative is still out there.
Pattern recognition across decades of institutional decisions is exactly the kind of work Clutch Justice was built to do. If you have a case, a set of documents, or a situation that doesn’t add up, a forensic record review maps the contradictions.
Consulting Tracks ?The Mixer Rupture
Jane Mixer was a 23-year-old University of Michigan law student. Her body was found in March 1969 at a cemetery east of Ypsilanti, clothed, covered with her raincoat, a copy of “Catch-22” placed by her side. She had not been sexually assaulted. She had been shot and strangled. She had posted a note on a campus bulletin board seeking a ride home to Muskegon. Someone answered it.
For 35 years, the task force’s position was that Collins killed Mixer along with the others. The physical differences in the case were acknowledged and then explained away as variations in method. Collins was convicted of the Beineman murder. The Mixer case was effectively closed by association.
In 2002, the Michigan State Police cold case unit submitted evidence from the Mixer case to the MSP forensic laboratory for DNA analysis. The results were not what the 35-year consensus predicted. The DNA profile developed from perspiration stains on a stocking found at the scene matched Gary Earl Leiterman, a retired nurse from Hamburg Township who had been arrested in 2001 for forging a prescription. His DNA had been entered into the state’s CODIS database following that arrest.
Collins was excluded entirely.
Then the laboratory found something else. A spot of blood from Mixer’s left hand matched the DNA profile of John Ruelas, a man who had been convicted of killing his mother in 2002. Ruelas’s DNA had also entered the CODIS database. In 1969, when Jane Mixer was murdered, John Ruelas was four years old and living in Detroit.
The MSP laboratory analyzed both the Mixer cold case evidence and Ruelas’s convicted offender sample in overlapping time windows. Leiterman’s defense argued that his DNA reached the Mixer evidence slides through cross-contamination during laboratory processing. Three scientists, including a UCSD professor who consulted with six DNA experts, subsequently published a statistical analysis concluding contamination was the most probable explanation for both DNA profiles. Leiterman was convicted in 2005 after fewer than three hours of jury deliberation. He died in prison in 2019 maintaining his innocence. The MSP laboratory’s response to the contamination question was to oppose the release of raw electronic data to outside analysts, with the deputy director characterizing external analysis as “tantamount to evidence tampering.” The contamination question was never independently resolved.
The Leiterman conviction is forensically contested. What is not contested is that the Michigan State Police laboratory produced results that included a four-year-old child’s DNA on 1969 evidence slides. The laboratory’s own explanation was that officers at the original crime scene had handled evidence without gloves and that some items had been lost over the intervening decades. Officers who worked the original Mixer scene confirmed at the Leiterman trial that evidence handling in 1969 was not what modern standards require. The biological evidence had degraded, been contaminated, and in some cases was simply gone.
The Lab Is the Constant
The Michigan State Police forensic laboratory is not a background detail in either of these failures. It is the variable that connects them across 35 years.
In 1970, the laboratory produced the NAA analysis that Conlin admitted into the Collins trial record and that the jury treated as scientific proof. In 2002 and 2003, the laboratory handled the cold case evidence in the Mixer matter under conditions that three scientists later concluded produced contaminated results. Between those points, the laboratory’s accreditation from the American Society of Crime Laboratory Directors lapsed and required multiple extensions from the national board, with the board warning state officials not to request another. The backlog of cases awaiting forensic analysis exceeded 19,000 in 2010, before marijuana legalization reduced the testing load. Over 11,000 untested rape kits were inherited from the Detroit Police Department crime lab when that facility closed in 2008.
Catherine Broad, whose brother Tim King was the fourth victim of the Oakland County Child Killer, has been the most persistent public voice documenting MSP’s cold case handling. She has documented the laboratory’s refusal to use third-party laboratories for evidence reanalysis, its refusal to upgrade DNA sequencing on the OCCK evidence, and its refusal to digitize the OCCK case files. Her documented comparison is pointed: the Boulder, Colorado Police Department, which was widely criticized for mishandling the JonBenet Ramsey investigation, has digitized all of its documents in that case, inventoried 2,500 pieces of evidence, and reviewed 40,000 reports. The Michigan State Police have not digitized the OCCK files.
A forensic scientist from the MSP Bridgeport crime lab described, in a 2024 interview, pulling out a pair of blue jeans from a cold case and watching them crumble when touched. The biological evidence in these cases does not wait. Storage conditions determine what science can later recover. The MSP’s storage practices in the late 1960s were, by the testimony of its own officers at the Leiterman trial, inadequate. Whether they remain so is a question the MSP has not been willing to answer in public.
What the Record Has Not Closed
Collins remains in prison. He has maintained his innocence for over 55 years. The conviction rests on forensic methods that no longer hold scientific validity, testimony from a cooperating witness who had his own legal exposure, and a prosecutor’s strategic decision to limit the scope of prosecution to the case with the strongest available evidence. The strongest available evidence was built on science that has since been repudiated by the Department of Justice and abandoned by the FBI.
This series is not a Collins innocence project. The evidentiary picture is genuinely complicated, and the forensic problems with the conviction do not automatically mean the verdict was wrong. What they mean is that the conviction cannot be evaluated on its merits because the scientific foundation it was built on no longer has any.
What this series is: a record-based examination of how seven murders produced one trial, how that trial was built on discredited science, how the public understanding of those murders was shaped by a journalist who trusted a prosecution that was itself operating on a contaminated foundation, how the laboratory that produced the original forensic work later contaminated the evidence that was supposed to answer the remaining questions, and how the men who ran these prosecutions and trials went on to long, honored careers in Michigan law.
There is one more thing to name. The degradation of the physical evidence is not a neutral misfortune. Because the biological samples were consumed by tests that produced worthless results, and because what remained was stored in conditions that contaminated it, the record is now effectively weaponized against any modern attempt at truth. You cannot reopen what you cannot test. The institution that produced the original failures benefits from the degradation of its own record. That is not a conspiracy. It is a structural incentive. And it is exactly what accountability journalism exists to document, even when the documentation is all that is left.
The record keeps asking questions. The institutions stopped listening. Clutch Justice has not.
Neutron Activation Analysis of hair samples and microscopic hair comparison. Both have been discredited. The FBI abandoned NAA for hair identification in the 1970s. A 2015 DOJ review found flawed hair comparison testimony in 90 percent of reviewed cases.
Prosecutor Delhey chose the case with the strongest available evidence. The Beineman prosecution was already the most expensive criminal proceeding in Washtenaw County history. Six other families received no prosecution.
Collins was excluded. Gary Leiterman matched. A four-year-old child’s DNA also appeared on the evidence, which three scientists later concluded indicated laboratory contamination. Leiterman was convicted in 2005 and died in prison in 2019.
Keyes himself noted that the Mixer case differed materially from the others but folded it into the Collins narrative because investigators said she belonged there. He trusted law enforcement’s synthesis over the evidentiary anomalies his own reporting had identified. The 2005 DNA result proved his instinct was correct and his conclusion was wrong.
Part Two examines the post-trial career trajectories of Judge Conlin and Prosecutor Delhey, the institutional culture that protected both, and what that culture produced. Part Three documents the men who had direct contact with the Busch/OCCK matter and the federal positions they subsequently held.
If you have documents and a situation that doesn’t add up, a forensic record review maps the contradictions, identifies the gaps, and produces a written findings memo you can act on — in 24 hours or less.
