The Molecule That Waited

A lab report is the shortest document in most homicide files and usually the most expensive. It rarely runs past three pages. The header gives the laboratory's name, an accreditation number, the requesting agency, and the agency case number. Below that comes a table: item number, item description, examination performed, result. Then a conclusions paragraph written in the careful passive voice of people who expect to be cross-examined. At the bottom are a signature, a title, and a date.

What follows describes the kind of report that sits in a file like April Tinsley's. I have not reproduced her laboratory's paperwork, and nothing here should be read as a quotation from it. The structure, though, is standard, and the facts of her case are public record.

April was eight. She lived in Fort Wayne, Indiana. On Good Friday, April 1, 1988, she walked a short distance through her own neighborhood to a friend's house to fetch an umbrella she had left there. She did not come home. Three days later she was found in a ditch in rural DeKalb County, roughly twenty miles away. She had been sexually assaulted and killed. That is as much of the injury as the investigation requires you to know, and more than her family should ever have had to read.

Look at the table on the report. One row describes an item of her clothing, collected, sealed, and logged in 1988. In the results column, the analyst records that biological material was found and that a DNA profile was developed. In the conclusions paragraph, the profile is described as originating from an unknown male. It was entered into CODIS, the FBI's Combined DNA Index System, which compares crime-scene profiles against those of convicted offenders, arrestees, and other cases. There was no match.

That is what it says. Here is what it misses. A CODIS profile looks at a small set of locations on the genome. Those markers are very good at telling you whether two samples came from the same person and nearly useless for telling you who that person is. A profile with no match is an exact description of a stranger. The question it forces is the one this case carried for thirty years: if the evidence already holds the answer, why can't anyone read it?

In the first chapter we looked at a report with a field left empty. Every case file has its own version of that line, the box where a name belongs. In April's file, that line stayed blank from 1988 into the summer of 2018.

The blank was not quiet. In May 1990, someone wrote on the door of a barn in Grabill, a small town near Fort Wayne, claiming to have killed her and threatening to kill again. In 2004, notes were left on girls' bicycles and at homes in the area, again claiming the killing and again threatening. Some were left with used condoms. Investigators collected these items, and the DNA on them matched the profile from April's clothing. The person who killed her was still nearby, still writing, and still unknown to every database that police could search.

Now consider what this did to the people working the case and to the families who lived in the neighborhoods where the notes appeared. A match without a name confirms the danger and offers nothing to act on. Tips came in for years. Composite sketches circulated. Men were looked at and cleared. None of them should be named here, and they will not be. The profile kept doing its one job: it ruled people out.

Then the method changed.

Forensic genetic genealogy depends on a different kind of DNA test. In place of the small set of CODIS markers, it reads hundreds of thousands of single-letter variations scattered across the genome. Together these make up a SNP profile, a pattern of single nucleotide polymorphisms, which are the same kind of data that consumer ancestry companies produce from a cheek swab. Because relatives share long stretches of DNA in predictable amounts, a SNP profile can be compared with others in a database, and the comparison can estimate whether two people are cousins, and roughly how distant.

From there the work is ordinary, slow, and largely done on paper. A genealogist takes the matches, which are usually distant cousins, and builds their family trees backward from public records: birth and death notices, obituaries, marriage licenses, census entries. She looks for the ancestors those cousins share. Then she builds forward again from those couples, through their children and grandchildren, looking for a man of the right age who lived in the right place. Nothing in a genealogy identification is a match in the courtroom sense. It produces a lead. The lead still has to be confirmed by a conventional DNA comparison.

In 2018 the Fort Wayne Police Department worked with Parabon NanoLabs, a private company in Virginia, whose genetic genealogy unit was led by CeCe Moore. The material from the case was used to generate a SNP profile. That profile was uploaded to GEDmatch, a free public site where people who had tested with consumer companies could upload their raw data and compare it with others. The matches pointed toward a family, and the tree narrowed toward two brothers living in the area. One of them was John D. Miller, of Grabill.

None of this would have been possible if the clothing had been lost, discarded, or allowed to degrade. Chapter Four described the evidence locker as a place where time stops, a room whose only task is to keep things exactly as they were when they came in. That description is accurate only when someone keeps doing the dull work of keeping it accurate. Every year, an item sealed in 1988 had to survive purges of old property, moves between storage facilities, retirements, budget cuts, and the reasonable-sounding argument that a case this old would never be solved. Biological evidence has to be dried, packaged, and kept away from heat and damp. The swab or cutting that produced the SNP profile in 2018 came from material someone had protected for three decades without any guarantee that it would matter.

The same technique raised serious questions, and the people who use it have not settled them.

The people who uploaded their DNA to GEDmatch were mostly hobbyists looking for great-grandparents or birth parents. They had not been asked whether their data could help police build a family tree that led to a relative. In April 2018, a few months before Miller's arrest, investigators in California announced an arrest in the Golden State Killer case, and the public learned how such identifications were being made. Critics asked fair questions. Should a person's genome be searchable by police because a cousin chose to upload theirs? What protects the innocent relatives who appear in the tree and are investigated along the way? In 2019 GEDmatch changed its default settings so that users had to opt in to law enforcement comparisons. Later that year the U.S. Department of Justice issued an interim policy limiting federal use of the technique mainly to violent crimes, and generally only after conventional methods have been tried. Some states have passed laws of their own. The debate continues, and a book that teaches this method should not pretend otherwise.

The tree pointed to two brothers. It did not say which one.

At this point investigators had two choices. They could approach the family, which risked warning the man they were looking for. Or they could find a way to get his DNA without his knowledge, which the law permits for items a person has thrown away. They chose the second. Officers watched Miller's trailer and collected his trash. Among it were used condoms. Readers who remember the 2004 notes will understand why that detail was significant.

The trash went to the lab. The analyst ran the standard comparison, the same kind of test that had been run on April's clothing three decades earlier. The report that came back is the one this chapter opened with, a newer version of the same three-page form. The table lists the discarded items. The results column records a profile. The conclusions paragraph, still in the passive voice, states that the profile from the discarded items is consistent with the profile developed from the evidence in the case, along with the statistical weight that such a statement carries.

On July 15, 2018, officers went to Miller's home in Grabill. According to court records and police statements, he admitted to the killing when they questioned him. A DNA sample taken after his arrest confirmed the match directly.

Go back to the first page of the file and the line that was empty for thirty years. It now holds a name, a date of birth, and an address about twenty miles from where April was found.

In December 2018, Miller pleaded guilty to murder and child molestation. He was sentenced to eighty years in prison. Because he pleaded guilty, there was no trial, no jury, and no public testimony in which the reasons were examined line by line. The docket records what happened in court. It does not record why it happened.

Read the newer lab report again, with the same three questions. It says the two profiles are consistent. It misses everything that made the result possible. It does not record the property clerk who kept the item through 1996 and 2005 and 2012. It does not record the decision to send material from a decades-old case to a private company when the method was still new and its legal footing was still being argued. It does not record the genealogist building a family backward from strangers' obituaries, or the officers waiting for trash day. A lab report records chemistry. The people who made the chemistry possible appear in it, at most, as item numbers.

The question it forces is about your own case, and about cases like it. Somewhere tonight there is a locker holding something sealed and labeled in another decade. Whether it can still speak depends on decisions being made now. Someone will choose whether old evidence is retained or destroyed. Someone will choose whether a degraded sample is used on a newer test or held back for a better one. Someone will decide whether a family is told that a request has gone out.

Those decisions have answers you can ask about. If you care about an unsolved child homicide in your community, and particularly if you are tempted to research it online, you do not need a suspect or a theory. Ask the investigating agency, in writing and politely, whether the biological evidence in the case has been preserved and inventoried. Ask whether the case has been reviewed for forensic genetic genealogy. Ask whether your state has an evidence-retention law for unsolved homicides, and what it requires. Ask the family, if they want to be asked, only what they have chosen to make public. If you have taken a consumer DNA test, read the law enforcement settings on any site that holds your data, and decide deliberately whether to opt in. Do not decide by default, in either direction.

Those requests take a letter and a stamp. They produce paperwork, which is the only thing that has ever kept evidence alive long enough to be read. April's family waited thirty years for the line on that first page to be filled. The swab that filled it had been sitting on a shelf the entire time, labeled in a stranger's handwriting, ready whenever someone found a way to read it.