The Uninvestigated Six: What Prosecutor William Delhey Chose Not to Prove
By Rita Williams | Clutch Justice | June 26, 2026
- Article 01: The Foundation Was Junk
- Article 02: The Uninvestigated Six (current)
- Article 03: Coming Soon
- Article 04: Coming Soon
William F. Delhey, Washtenaw County Prosecutor, tried John Norman Collins for one murder: Karen Sue Beineman, age 18, killed in July 1969. Collins was convicted and sentenced to life in prison. Six additional murders, attributed to Collins by MSP investigators, were never brought to trial. The decision to prosecute a single case was tactically defensible. It was also institutionally consequential. Fifty-five years later, those six cases have no trial record, no adjudicated evidentiary findings, and no formal record of prosecution.
The One Case That Went to Trial
The Karen Sue Beineman case was the strongest available against Collins for a specific and documentable reason: physical evidence recovered from a controlled location.
Beineman, 18, an Eastern Michigan University student, disappeared on July 23, 1969. Her body was found three days later. MSP investigators placed her at a motorcycle shop associated with Collins shortly before her disappearance. The cause of death was blunt force trauma and strangulation.
The prosecution’s central forensic exhibit was hair. Human hair clippings recovered from the basement of Collins’ uncle, MSP Corporal Leroy Matheson, were found in the victim’s undergarment. Forensic analysts testified the clippings were consistent with a haircut given in that basement. The basement was Collins’ domain during the weeks before Beineman’s disappearance.
This was the most direct physical connection the prosecution held to any of the seven deaths attributed to Collins. The other six cases depended on pattern, geography, victimology, and circumstantial timelines. The Beineman case had an evidentiary anchor tied to a specific location under Collins’ control.
Delhey tried the Beineman case. The jury returned a guilty verdict in August 1970. Judge Conlin imposed a life sentence. Collins has been incarcerated since that verdict.
The Forensic Profile of the Remaining Six
The six cases Delhey did not try shared a common evidentiary condition: the physical evidence connecting Collins to each victim was weaker, more degraded, or more dependent on the contested forensic methods documented in Article 01 of this series.
The earliest attributed victim. Fleszar was found more than two years before Collins’ arrest. By the time investigators focused on Collins in 1969, the physical evidence in her case had aged past the point where a direct forensic connection to a specific suspect could be established on the available record.
Schell was last seen June 30, 1968. Witnesses placed her in a vehicle with men, one resembling Collins. Eyewitness identification under those conditions carried the standard evidentiary vulnerabilities. No physical evidence directly connecting Collins to Schell’s death was available at trial-quality standard.
Mixer was shot twice with a .22-caliber handgun and strangled. The cause of death and forensic profile did not match the dominant pattern seen in the other attributed cases. Investigators included her in the cluster, but the evidentiary basis for Collins’ connection was limited. In 2002, DNA testing on physical evidence from the Mixer case identified John Ruelas, a man with no known connection to the Michigan Murders investigation. Ruelas was convicted of Mixer’s murder in 2005 by Washtenaw County Circuit Court.
Skelton exhibited the victimology and geographic pattern MSP attributed to Collins. No physical evidence directly connecting Collins was available at trial-quality standard on the investigative record.
The youngest attributed victim. Basom was strangled. The case fit the investigative pattern but produced no direct physical link to Collins that met the evidentiary threshold Delhey applied in selecting the Beineman case for prosecution.
Kalom was shot and stabbed. The firearm used was never recovered. Like Schell, she was shot, which differed from the strangulation-dominant pattern in the other attributed cases. The firearms evidence was absent from the investigative record.
Taken together, the six cases presented an evidentiary constellation that investigators found persuasive as a pattern but that a prosecution would need to establish case by case, victim by victim, against the individualized burden of proof at trial.
The Clutch Justice Field Kit courses walk through the institutional mechanics of prosecutorial conduct, court system patterns, and accountability documentation. Built for families, researchers, and advocates working with the actual record.
Browse the Courses ?The Prosecutorial Calculus
The institutional logic of Delhey’s decision can be stated structurally without conjecture.
A life sentence is the maximum custodial result achievable under Michigan law for first-degree murder. Collins received that result from the Beineman conviction. From a custodial standpoint, additional convictions would have produced no different outcome. Collins was not leaving prison.
The risk calculation ran in the opposite direction. Trying Collins on weaker evidentiary cases created the possibility of acquittal. An acquittal in a high-profile case does institutional damage. It generates appellate ammunition. It seeds reasonable doubt in the public record. For a prosecution that had secured a life sentence, the incentive to assume further evidentiary risk was structurally low. The strategy was rational. The cost was borne entirely by the six cases that were not tried.
This pattern, prosecuting the strongest case to secure the custodial result and allowing remaining cases to close without trial, is a documentable institutional dynamic in major homicide prosecutions. It is not unique to this case or this prosecutor. It is the predictable output of a system that optimizes for conviction rate and custodial certainty rather than full evidentiary development on the record.
The decision to try one case instead of seven was not a failure of investigation. It was a successful application of the risk management framework that governs prosecutorial charging decisions. The failure embedded in the record is the absence of any parallel mechanism to account for the cases the prosecution chose not to build.
What the Uninvestigated Six Produced
Six trials did not happen. What that means institutionally is specific.
No trial record was created for the deaths of Mary Fleszar, Joan Schell, Jane Mixer, Maralynn Skelton, Dawn Basom, or Alice Elizabeth Kalom in connection with John Norman Collins. No evidence was formally introduced and tested through cross-examination. No witness testified under adversarial conditions. No judicial ruling established the admissibility or inadmissibility of any physical evidence associated with those deaths. No verdict of any kind was recorded.
The six cases exist in a state that is technically open and functionally closed. MSP attributed the murders to Collins. Collins received a life sentence. The institutional incentive to revisit six uninvestigated murders with a decades-old evidentiary record and a defendant already serving life is structurally minimal.
The Jane Mixer case demonstrates what is embedded in the uninvestigated six. In 2002, MSP’s own laboratory identified DNA evidence pointing to John Ruelas, a man with no prior connection to the Michigan Murders investigation. Ruelas was convicted of Mixer’s murder in 2005. The Collins attribution was effectively set aside without a trial record to anchor or complicate it. That result, the posthumous fracturing of an attribution that was never tested in court, was structurally possible precisely because there was no trial record. The other five uninvestigated attributions carry the same latent instability.
The Record the Prosecution Did Not Make
The Michigan Murders of 1967 to 1969 are documented in the record that exists: the 1970 trial transcript, the forensic exhibits, the Collins conviction. They are also defined by the record that does not exist.
What Delhey chose not to prove is inaccessible through the trial record because there is no trial record on those six cases. The evidentiary basis for attributing each of the six additional murders to Collins was never subjected to adversarial testing. Investigators believed Collins was responsible. That belief was never tested in court. The families of six victims received no verdict in connection with Collins. The archive contains no adjudicated findings on six attributed murders.
That gap is a record the prosecution built by deciding not to build it. The choice to try one case was also, structurally, a choice not to try six. The institutional consequence is a fifty-five-year archive in which six attributions exist without evidentiary adjudication, six cases are closed without verdicts, and six evidentiary records degraded in the absence of the trial process that would have stabilized them.
Article 03 of this series examines the institutional pipeline that shaped the Michigan Murders record from the outside: the prosecutors, federal officials, and judicial actors whose documented decisions determined what the archive would and would not contain.
Delhey tried the Karen Beineman case because it held the strongest available forensic evidence: hair clippings from Collins’ uncle’s basement found in the victim’s undergarment. A single conviction carried a life sentence, making additional prosecutions on weaker evidentiary grounds an institutional risk with no custodial upside.
The six murders attributed to Collins were never prosecuted. They remain technically open with no verdict and no trial record. They were effectively closed when Collins received his life sentence for the Beineman murder in 1970.
The 2002 DNA identification of John Ruelas, and his 2005 Washtenaw County conviction for Mixer’s murder, showed that one of the six attributed cases did not involve Collins. Whether the remaining five attributions are stable is an open evidentiary question. Without trial records, there is no judicial framework to assess or challenge them.
No charges have been filed. The physical evidence is more than five decades old, Collins has been incarcerated since 1970, and the evidentiary barriers are significant. No public record indicates any active prosecution effort on the remaining attributed cases.
- People v. Collins, Washtenaw County Circuit Court (1970). Trial verdict and sentencing. Judge John W. Conlin presiding.
- People v. Ruelas, Washtenaw County Circuit Court (2005). Conviction for the murder of Jane Mixer following DNA identification.
- Michigan State Police. Michigan Murders Task Force investigative files (1967-1969). Attribution of seven murders referenced in trial exhibits and subsequent published accounts.
- MSP Forensic Science Division. DNA analysis, Jane Mixer case (2002). Contributor profile identified as John Ruelas; subsequent conviction 2005.
- Keyes, Edward. The Michigan Murders. Reader’s Digest Press, 1976. Primary published account of the investigation and prosecution; reflects MSP investigative attribution of seven cases to Collins.
Continue Your Investigation
If this reporting raised more questions, use the Clutch Justice ecosystem to keep going.
