by Richard Resch

The Gold Standard That Isn’t

On the night of November 29, 2012, a group of men broke into a mansion in Monte Sereno, California, and killed its owner, Raveesh Kumra, a 66-year-old Silicon Valley investor. When the county crime lab processed evidence from the scene, a DNA profile recovered from Kumra’s fingernail clippings produced a hit in California’s felony database. The match pointed to Lukis Anderson, a 26-year-old homeless man who panhandled on the streets of San Jose, 10 miles away.

Anderson was charged with first-degree murder. The charge carried a possible death sentence. His court-appointed attorney, Kelley Kulick, sat down with him at the Santa Clara County jail. Anderson couldn’t explain how his genetic material ended up under a dead man’s fingernails. He drank heavily. He blacked out. He told Kulick that maybe he’d done something he couldn’t remember. Kulick told him to stop talking and let her investigate the evidence.

What she found changed the case entirely. Anderson’s medical records showed he had been picked up off a sidewalk by paramedics that same evening, severely intoxicated and nearly comatose. He was admitted to a hospital and remained under continuous medical supervision through the night. Staff records documented his presence during the precise hours when Kumra was being bound, robbed, and suffocated with duct tape.

The explanation, once investigators pieced it together, was as simple as it was alarming. The same two paramedics who transported Anderson to the hospital earlier that evening responded to the Kumra mansion about three hours later and checked Raveesh Kumra’s vitals. Prosecutors, the defense, and police agreed that the paramedics somehow moved Anderson’s DNA from San Jose to Monte Sereno, but they could not determine the precise mechanism. Santa Clara County District Attorney Jeff Rosen suggested a pulse oximeter might have been responsible. Kulick thought uniforms or other equipment were just as plausible. Either way, the point was the same. Routine emergency-response contact appears to have linked an innocent man to a killing he could not have committed.

Anderson spent roughly five months in jail before the charges were dismissed. After his release, he returned to the streets and received no compensation for his time in jail. The Kumra prosecution nevertheless moved forward against other suspects. DeAngelo Austin and Javier Garcia were later convicted, and a third man, Marcellous Drummer, was also prosecuted after his DNA was later found on evidence from the scene. In Anderson’s case, the DNA hit had pointed investigators to the wrong man.

From Bloodstains to Skin Cells

Anderson’s case is extreme, but it is not an anomaly. It illustrates a problem that runs through the modern forensic system, i.e., the gap between what DNA evidence actually proves and what police, prosecutors, judges, and jurors assume it proves. In the public imagination, a DNA match operates like a molecular fingerprint stamped at the scene of the crime. And for certain categories of evidence, that perception contains a kernel of accuracy. When investigators recover semen from a sexual assault or blood spatter from a violent struggle, the biological material is often directly connected to the criminal act. The statistical power of a full DNA profile, which can narrow the odds of a coincidental match to one in billions or more, gives these identifications extraordinary weight.

But the Anderson case exposes a fault line that has widened as forensic technology has grown more sensitive. In 1997, Australian forensic scientist Roland van Oorschot and his colleague Maxwell Jones published a brief paper in Nature demonstrating that usable DNA profiles could be generated from the skin cells people leave behind when they touch objects. Roland A.H. van Oorschot & Maxwell K. Jones, DNA Fingerprints from Fingerprints, 387 Nature 767 (1997). Although the paper was only nine paragraphs long, its implications have been unfolding ever since.

Before that discovery, crime labs generally needed visible biological material to work with, e.g., a bloodstain, a semen sample, saliva on a cigarette butt. The van Oorschot findings opened the door to what became known as “touch DNA,” and the technology for recovering and amplifying these vanishingly small deposits has only grown more powerful. Commercial kits can now generate a complete genetic profile from as few as three to five cells. In Harris County, Texas, touch DNA case submissions more than tripled between 2009 and 2013. DNA evidence now appears in categories of crime – burglaries, thefts, weapons possession – where it was once unthinkable. And the smaller the sample, the larger the interpretive problems become.

The central difficulty is that DNA can identify who left biological material, often with astonishing precision, but in a growing number of cases, DNA cannot answer the question that actually matters at trial, which is how and when that material got there. Kumra’s fingernail clippings yielded DNA matching Anderson. That was a verified forensic result. The inference that Anderson had struggled with Kumra during a home invasion was not a scientific conclusion at all. It was a story constructed around a data point, and it was completely wrong.

What This Article Will Show

This distinction between identification and explanation is the central concern of this article. It will examine how the criminal legal system’s most trusted form of physical evidence can mislead when its limitations are not understood, not disclosed, or not taken seriously. Cells travel. They migrate through handshakes, shared surfaces, household pets, air and dust, and the gloves and instruments of the people collecting evidence. DNA can arrive at a crime scene hours or days before the crime occurs or be deposited long after by investigators themselves. It carries no timestamp. It encodes no narrative. And the sophisticated computer programs now used to interpret the most complex DNA mixtures often operate behind proprietary code, forcing defendants to litigate whether they may inspect the source code or related materials and, in some cases, leaving them with no access or only restricted access under protective-order conditions.

None of this means DNA evidence is worthless. Post-conviction DNA testing was central to 375 exonerations nationwide through 2020, according to the Innocence Project’s historical count. Twenty-one of those 375 people had served time on death row. DNA remains one of the most powerful tools the justice system has, both for identifying the guilty and for freeing the innocent. But that power depends on honesty about the tool’s limits.

The pages that follow will document real cases in which real people lost years or decades of their lives because the system treated a biological trace as irrefutable proof of guilt. And they will identify concrete reforms, drawn from the best available science and the recommendations of the field’s own governing institutions, that would bring courtroom practice closer to scientific reality.

Lukis Anderson was lucky. He had an alibi no one could dispute, documented minute by minute by hospital staff. Most people whose DNA ends up where it inexplicably shouldn’t be are not so fortunate.

What Touch DNA Is and What It Isn’t

Humans are prolific shedders. The average person discards roughly 50 million skin cells every day, leaving invisible biological traces on virtually every surface they contact. Those traces contain DNA. And in the nearly three decades since Roland van Oorschot and Maxwell Jones demonstrated that usable genetic profiles could be extracted from objects people had merely touched, forensic laboratories worldwide have built an expanding practice around recovering and analyzing these minuscule deposits. The field calls them “touch DNA,” “contact DNA,” or “trace DNA.” The terms are used interchangeably, but none of them is quite accurate. The imprecision is more important than most people realize.

When a person handles an object, the biological material left behind is not simply a cluster of shed skin cells sitting neatly on the surface. The deposit may include intact or degraded keratinocytes, nucleated epithelial cells from saliva or perspiration already present on the person’s hands, and free-floating fragments of cell-free DNA. A 2026 review in the Egyptian Journal of Forensic Sciences noted that considerable scientific debate persists about the exact composition of touch DNA deposits and that cell-free DNA has in some studies produced higher yields than intact cells. The practical implication is that when an analyst recovers DNA from a doorknob or a weapon, the genetic material may not have come from the skin cells of the last person who gripped it. It could have come from saliva transferred when someone wiped their mouth or from someone else’s cells already sitting on the handler’s palm.

The quantities involved are staggeringly small. Forensic laboratories routinely attempt to develop profiles from fewer than 100 picograms of DNA, an amount equivalent to roughly 15 to 20 human cells. Some commercial amplification kits can generate results from as few as three to five cells. Working at these thresholds means analysts are often operating at the edge of their instruments’ detection capabilities, where the difference between a genuine signal and background noise can be indistinguishable.

That sensitivity is both an achievement and a liability. When crime labs could only extract DNA from visible biological stains, the evidentiary link between the sample and the crime was usually straightforward. Touch DNA changed the equation. A profile can now be developed from a steering wheel, a light switch, or the grip of a firearm. DNA evidence now appears in categories of crime – burglaries, thefts, weapons possession – where it was once unthinkable.

The Shedder Problem and the Limits of Collection

The smaller the sample, the more fragile the results. A 2022 systematic review led by Pamela Tozzo at the University of Padova compared single-swab, double-swab, and other collection methods and concluded that the literature did not establish a universally superior sampling technique. In the review’s own descriptive analysis, the single-swab method showed higher efficiency in DNA recovery across a wide variety of experimental settings, while results still varied by surface type, environmental conditions, and the characteristics of the person who left the deposit. One of the most important variables the review identified is what researchers call “shedder status.” Some people deposit substantially more DNA through casual contact than others, and this propensity can vary within the same individual at different times, depending on hand moisture, recent hand-washing, and what the person touched beforehand.

The implications for criminal cases are significant. If a “good shedder” recently shook hands with a “poor shedder” who then handled a weapon, the weapon may yield a stronger profile from the person who never touched it than from the person who did. Cynthia Cale and her colleagues at the University of Indianapolis demonstrated precisely this in a 2016 study published in the Journal of Forensic Sciences. Participants shook hands for two minutes, then immediately handled knives cleaned of prior DNA. Secondary DNA transfer was detected in 85 percent of the samples. In five of those samples, the secondary contributor, the person who never handled the knife, was either the only identifiable contributor or the major one.

Cale’s results go to the heart of the term “touch DNA” itself. The phrase implies direct physical contact. In police reports and courtroom testimony, saying that touch DNA was found on an object carries an implicit suggestion that the identified person touched that object. But the science does not support that inference as a general proposition. DNA can arrive on a surface through an intermediary, through environmental contamination, or through the actions of evidence collectors. The term describes a type of biological deposit, not a mechanism of deposition.

What Gets Lost in the Lab

Laboratory processing compounds the uncertainty. The standard forensic workflow of extraction, quantification, and amplification can result in the loss of genetic material at multiple stages. For touch DNA deposits, which start with almost nothing, that attrition can be consequential. Researchers funded by the National Institute of Justice (“NIJ”) have explored alternatives like direct polymerase chain reaction (“PCR”), which bypasses extraction and pre-amplification quantification, and have reported better results on some substrates but worse results on others. As of July 1, 2025, the FBI’s Quality Assurance Standards (“QAS”) no longer categorically require pre-amplification quantification for every forensic sample. For laboratory-defined forensic sample types, a laboratory may use a validated method that calculates DNA quantity after or during amplification. Direct PCR therefore remains a qualified, substrate-dependent option under current federal standards when validated for the relevant forensic sample type and performed in compliance with the QAS requirements.

Importantly, touch DNA is not a single, well-defined category of evidence with predictable behavior. It is a catchall term for genetic material recovered in extremely small amounts from surfaces that may or may not have been directly touched by the person whose profile is eventually identified. Its composition is variable. Its deposition mechanism is often unknown. Its recovery depends on technique, surface type, and the biological idiosyncrasies of the individuals involved. None of this is secret or contested among forensic scientists. The problem is that the knowledge has not reliably made its way into the courtroom, where the phrase “touch DNA” still carries an air of certainty that the science behind it does not support.

The Core Mistake: DNA Can Identify but Often Cannot Explain

Forensic scientists who work with DNA evidence often analyze it through what they call a hierarchy of propositions. The framework has been developed over decades in the literature, formalized in international guidance, and incorporated into current regulatory guidance. Current U.K. Forensic Science Regulator guidance identifies four levels of issue in forensic practice and criminal proceedings: (1) offence, (2) activity, (3) source, and (4) sub-source.

For the purposes of this article, the most important distinctions are among levels 2, 3, and 4 because confusion among them can make a powerful identification statistic appear to answer a question it was never designed to answer. At the sub-source level, the question is statistical: what is the probability that the recovered profile originated from a particular person rather than from someone else in the population? This is where the dazzling numbers live, making matches with statistical probabilities of one in several billion or one in several trillion. These figures are genuinely impressive, but they answer only a narrow question. They tell you whether a particular person could be the source of the DNA. They say nothing about how it arrived where it was found.

The source level asks whether the person of interest actually is the source, accounting for coincidental matches, laboratory error, or contamination. In many cases, the source of the DNA is not disputed at all. Defense attorneys sometimes concede their client’s DNA is on the item. The dispute is about what that identification means.

That brings us to the level that matters most at trial and receives the least rigorous treatment, viz., the activity level. Here, the question shifts from who left the DNA to how it got there and when. Did the defendant grip the knife during a stabbing, or did his skin cells transfer to the handle through a handshake? Did the defendant wear the gloves found at the crime scene, or had she handled them at a store weeks earlier? These are the questions that determine guilt or innocence. And they are precisely the questions that a DNA profile, standing alone, cannot answer.

The Framework the Field Built and America Declined to Adopt

The International Society for Forensic Genetics (“ISFG”) addressed this problem directly in a landmark 2020 guidance paper authored by Peter Gill, Tacha Hicks, John Butler, and 10 other leading forensic scientists. The paper is unequivocal. The statistical weight calculated for a DNA profile at the sub-source level cannot be carried over to the activity level. A match probability of 1 in 10 billion tells you the DNA almost certainly belongs to the defendant. It tells you nothing about whether the defendant committed the alleged act. The ISFG Commission warned that conflating the two risks a logical error that can distort the outcome of a trial.

Hicks and colleagues elaborated on this framework in a 2022 review published in Genes, emphasizing that evaluating DNA evidence at the activity level requires the forensic scientist to consider transfer, persistence, prevalence, and recovery, variables the literature abbreviates as TPPR. Without this analysis, a DNA report can present a statistically overwhelming match figure while leaving entirely unaddressed the question the jury most needs answered.

Consider a simplified example. A defendant is charged with assault. His DNA is found on the victim’s shirt collar. The prosecution says he grabbed the victim during the attack. The defense says he hugged the victim at a social gathering earlier that day. At the sub-source and source levels, the evidence is identical under both propositions. Only at the activity level does it acquire different meaning, and evaluating that meaning requires knowledge about how readily DNA transfers during gripping versus embracing and how long it persists on cotton fabric.

This kind of evaluation is standard practice in parts of Europe. The Netherlands, Switzerland, and the U.K. have incorporated activity-level assessment into their forensic reporting frameworks to varying degrees. The U.K. Forensic Science Regulator, in guidance updated as recently as March 2026, requires practitioners to consider the appropriate level of propositions for each case.

In the U.S., the picture is starkly different. Scientific Working Group on DNA Analysis Methods (“SWGDAM”), an FBI-affiliated working group that develops guidance documents and recommends revisions to the FBI Director’s QAS, published a December 2025 position statement identifying what it described as significant scientific, practical, and legal impediments to formal activity-level reporting in the U.S. A 2025 review by Janet Stacey and colleagues likewise described broader barriers to adoption, including methodological reticence, concern about the lack of robust and impartial data to inform probabilities, regional differences in regulatory frameworks and methodology, and the training and resources needed to implement these evaluations. Practitioners increasingly face activity-level questions on the witness stand even where no formal framework exists for answering them, leaving individual analysts to address issues the field has not implemented systematically.

The Void the Jury Fills

The result is a structural gap at the center of American DNA evidence. Laboratories report results at the source or sub-source level, presenting match statistics that convey extraordinary certainty about identity. Jurors then draw their own conclusions about activity, filling the interpretive void with assumptions shaped by popular culture, prosecutorial narrative, and the sheer persuasive weight of numbers they cannot contextualize. The forensic scientist may never have said the defendant gripped the murder weapon. But by presenting a match probability in the trillions and offering no framework for understanding how else the DNA might have arrived, the effect can be the same.

This is not a failure of individual prosecutors or analysts. It is a systemic design problem. The American forensic reporting system is built to answer the question “whose DNA is this?” with remarkable precision. It is not built to answer the question “how did it get there?” And in the categories of cases that increasingly dominate the forensic docket – cases involving touch DNA, low-template samples, mixed profiles, and indirect transfer – the second question is the crucial one.

The Invisible Hand: Secondary and Tertiary DNA Transfer

A person shakes your hand at a party. Immediately afterward, you pick up a kitchen knife. If police later collect that knife as evidence, the strongest profile on the handle could belong to the person who shook your hand. That person may never have entered your kitchen or touched the knife. Nevertheless, in the language of a forensic report, their DNA was “recovered from the weapon.”

The mechanics are not hypothetical. They mirror the experimental finding of the Cale study introduced earlier, one of the clearest demonstrations in the published literature of secondary DNA transfer. The concept is straightforward. Primary transfer occurs when a person’s biological material moves directly from their body to an object or another person. Secondary transfer occurs when that material travels through an intermediary, arriving at a destination the original donor never contacted. Tertiary transfer adds another link in the chain. Each step attenuates the signal, but modern amplification technology is sensitive enough to detect it.

Van Oorschot and Jones identified the possibility of secondary transfer in their foundational 1997 paper, finding alleles on vinyl gloves that did not belong to the wearers. In the years that followed, some researchers attempted to demonstrate that secondary transfer was too rare or too weak to produce meaningful forensic profiles. Those early reassurances have not held up.

What Two Decades of Research Have Confirmed

As amplification kits grew more sensitive through the 2000s and 2010s, the detection of indirectly transferred DNA became more frequent and more forensically significant. The Cale study detected secondary transfer in 85 percent of samples. In five of those samples, the person whose DNA traveled through the handshake was either the sole identifiable contributor or the dominant one. As the authors observed, presenting such results in court as evidence that the identified person had handled the weapon would be persuasive and, in those five cases, wrong.

A 2023 systematic review by Francesco Sessa and colleagues analyzed 49 published studies on indirect DNA transfer spanning more than two decades. The review confirmed that secondary transfer is a reproducible phenomenon influenced by multiple interacting variables such as shedder status, the type and duration of contact, the nature of the receiving surface, environmental conditions, and the sequence of contacts, which is difficult to reconstruct after the fact. The review supports the conclusion that indirect transfer can create forensic results that appear to place a person’s DNA on an item or in a setting the person never directly touched.

The van Oorschot group’s own 2021 review catalogued the pathways through which DNA can travel indirectly. Shared surfaces serve as transfer points. Clothing transfers DNA between wearers and objects. Hands accumulate DNA from every surface they contact and redistribute it to the next. Cleaning may not eliminate DNA. One study discussed in the review found that cleaning actions sometimes redistributed biological material rather than removing it and that chloric cleaning agents rendered almost everything DNA-free. Firearms handled in realistic scenarios also showed wide variability in recovered profiles, with indirect pathways remaining part of the interpretive problem.

Transfer Without a Crime, Deposits Without a Trace

The implications extend beyond handshakes and kitchen knives. Officers who touch a doorframe and then handle an exhibit with the same pair of gloves can migrate DNA from one place to another. Evidence items packaged together can exchange biological material within the same container. More broadly, reused or inadequately cleaned tools, benches, gloves, and other investigative surfaces can mediate transfer in ways unrelated to any criminal act by the person whose DNA is found. None of these transfers requires direct contact between the DNA donor and the object on which the DNA is later detected.

Van Oorschot and colleagues’ 2019 review also describes what researchers call “DNA parking.” The term does not refer simply to old DNA lingering on a surface. It refers to DNA that is secondarily transferred to an intermediate object, later picked up again, and then deposited on the surface of interest. That concept underscores how a recovered deposit can reflect a multi-step pathway rather than direct contact. Separately, DNA deposited through routine innocent contact can also persist and later be recovered after a crime, but that persistence problem is analytically distinct from DNA parking.

What makes this particularly dangerous is the difficulty of distinguishing primary from secondary transfer after the fact. In a 2017 study, Duncan Taylor and colleagues used Bayesian-network modeling to evaluate whether DNA quantity and related profile information could help distinguish primary from secondary transfer events. The model showed some ability to distinguish the two under the controlled conditions tested, but the authors stressed the limited dataset and the need for additional ground-truth testing. The practical point is that quantity and profile quality may inform the analysis, but they do not, standing alone, mechanically prove direct contact.

The forensic system’s vocabulary reinforces the confusion. As the Cale study explicitly warned, terms like “touch DNA” and “wearer DNA” carry an inherent implication of direct physical contact. When an analyst testifies that touch DNA matching the defendant was recovered from a gun grip, jurors hear that the defendant touched the gun. Criminal Legal News (“CLN”) addressed this gap in a 2018 article warning that prosecutors and DNA experts will almost never proactively explain the possibility of secondary DNA transfer to juries, even though jurors need to understand that DNA can be present for reasons other than direct contact.

Finding someone’s DNA on an object does not mean they touched it. The science has established this clearly. What remains largely unchanged is the courtroom presentation, where a DNA match is still routinely treated as though it answers the question of how the genetic material arrived.

DNA Without a Timestamp: Background DNA, Persistence, Collection, and Packaging

DNA carries no date stamp. When a forensic analyst recovers a genetic profile from a doorknob, a shirt collar, or the surface of a victim’s skin, the result does not, by itself, reveal when that biological material was deposited. It could have been left minutes before collection or weeks. It could predate the crime by days, or it could have arrived afterward through investigator contact, environmental drift, or the ordinary comings and goings of people who share a space. The profile itself is silent on that critical question. And yet the forensic system routinely treats recovered DNA as though it is contemporaneous with the alleged criminal act, even though no validated routine casework method can reliably determine when a trace deposit was made.

Persistence: How Long Does Touch DNA Last?

The most comprehensive study to date on environmental persistence of touch DNA was conducted by Meghan Ramsey’s group at MIT Lincoln Laboratory with National Institute of Justice funding. Researchers deposited touch DNA and control DNA onto stainless steel and cotton fabric and then exposed the samples to different combinations of temperature, humidity, and ultraviolet light. The clearest result was that UV exposure had the greatest effect on degradation. By contrast, in the absence of UV light, the touch-DNA samples on stainless steel were generally stable over the seven-day study period. The van Oorschot group’s 2021 review also cited separate work from Singapore that tracked DNA on a variety of substrates for up to 85 weeks and found that indoor samples were comparatively stable over long periods.

For the courtroom, the persistence data cuts in a troubling direction. If a defendant handled an item days or weeks before a crime, and the item was stored indoors, the defendant’s DNA can survive to be recovered and presented at trial as though it were evidence of the criminal act. The longer DNA persists, the wider the window of innocent explanation becomes.

Background DNA: What Was Already There

The problem is compounded by the pervasive presence of what researchers call background or prevalent DNA, i.e., genetic material that exists on objects and surfaces as a normal consequence of everyday life. A 2025 review by Cara Woollacott and colleagues at Flinders University, published in Forensic Sciences, examined the transfer, prevalence, persistence, and recovery of DNA from human body surfaces. The review documented that non-self DNA is routinely found on the skin of people who have had no criminal contact with anyone. Hands carry a constantly shifting mixture of self-DNA and DNA from other people and surfaces encountered throughout the day. The review further noted that the non-self DNA already present on a victim’s body before any crime occurs can complicate interpretation, because analysts may be unable to distinguish between DNA deposited during the criminal act and DNA that was there all along.

Consider a domestic assault case in which a victim’s neck is swabbed for the suspect’s DNA. If the suspect and victim are cohabitants, the suspect’s DNA may be prevalent on the victim’s skin as a background condition of shared living. Finding the suspect’s DNA on the victim’s neck does not, by itself, distinguish between a strangling and a hug that happened the previous day.

Collection and Packaging: Shaping the Evidence

The decisions investigators make about where and how to collect biological evidence are not neutral. Woollacott’s 2025 review noted that optimal recovery can depend on the body surface sampled and that interpretation should take account of which body areas are typically contacted during different activities, as well as the possibility that non-self DNA was already present there before the event in question. The swabbing technique itself also introduces variability. The DNA profile presented in court is therefore shaped not only by who contributed DNA, but by where investigators chose to sample and how they collected it.

Perhaps the least intuitive source of interpretive error is what happens after evidence leaves the crime scene. Multiple evidentiary items can be packaged together for transport to the laboratory, creating an opportunity for DNA to move from one item to another after collection. A 2025 study by Yong Sheng Lee and Christopher Syn at Singapore’s Health Sciences Authority investigated this directly. Items handled by different participants were stored together in the same evidence package for four to five days under passive conditions – no friction, no force, and no long-distance movement. DNA transfer between items within the package was observed on 39 percent of the studied items. For 10 percent of recovered samples, the transfer changed the source attribution of the DNA profile, meaning that DNA from one person’s item appeared on the other’s in sufficient quantity to alter the interpretive result.

Thirty-nine percent transfer under passive storage is a striking finding. It suggests that placing two exhibits in the same bag can introduce DNA from one onto the other, producing a result the laboratory may interpret as evidence of contact that never occurred.

Each of these factors – persistence, background prevalence, collection methodology, and packaging transfer – is individually significant. Together, they create a compounding uncertainty that standard forensic reporting does not address. The profile is real. The statistics confirming the identity match are genuine. But the narrative connecting that profile to the criminal act depends on assumptions the science cannot validate. The further the evidence moves from plentiful biological material with a clear deposit mechanism, into the realm of touch deposits and indirect pathways, the more the interpretive burden shifts from science to storytelling.

Silent Witnesses and Ambient Traces: Pets, Air, Dust, and Water

The previous sections examined how DNA travels through handshakes, shared surfaces, investigator equipment, and evidence packaging. The research surveyed here adds more pathways that complicate the picture even further. Human DNA can be redistributed by household pets, suspended in indoor air, accumulated in dust, and shed into water. Some of these routes depend on physical contact. Others do not. Together, they show how genetic material can move through an environment in ways that may have little to do with the criminal act investigators are trying to reconstruct.

Your Dog Knows Where Your DNA Has Been

Dogs and cats are among the most common household pets worldwide. They move through domestic spaces, rub against occupants and visitors, and interact physically with the people who enter their environments. From the standpoint of DNA transfer, that makes them potential intermediaries rather than passive background features.

A series of studies led by Heidi Monkman at Flinders University, in collaboration with Roland van Oorschot of the Victoria Police Forensic Services Department, has systematically documented this role. In a 2023 study published in Genes, Monkman and colleagues found that human DNA was prevalent on household dogs and that dogs could act as vectors for indirect transfer, including transfer from dogs to gloved hands during patting and to a sheet while walking. Later work reported comparable concerns for cats, including transfer from cats to gloved hands after brief contact.

The most forensically dramatic demonstration came in a 2026 study simulating a “dog-napping” scenario. Five dogs were placed into five unfamiliar cars by a recruited handler for 20 minutes. Dog owners’ DNA, carried on their animals’ fur, was detected inside cars the owners had never entered in 35 percent of samples. The handler’s DNA appeared on the dogs in 40 percent of cases. And the car owners’ DNA was found on the dogs in five percent of samples, despite the car owners having had no contact with the animals. DNA traveled from person to dog, from dog to car, from car to dog, and from dog to handler, a web of transfers generated by 20 minutes of ordinary interaction.

A 2026 review by Carla Bini and colleagues at the University of Bologna synthesized this emerging body of work, confirming that dogs and cats readily acquire human DNA following even brief contact; that the DNA can be redistributed to gloved hands, vehicles, and surfaces; and that higher-order transfer events have been documented. For the criminal legal system, the implications are direct. The experimental data do not prove what happened in any particular case, but they do document transfer pathways that make such alternative explanations scientifically plausible. A person who encounters a household dog may acquire the owner’s DNA from the animal, and the animal may later carry another person’s DNA to surfaces that person never touched. Those possibilities should be evaluated as activity-level explanations, not dismissed merely because a DNA profile appears in an unexpected place.

The Air You Breathe, the Dust You Leave

A human adult sheds approximately 1,000 skin cells per square centimeter of body surface per hour. These cells are small enough to become aerosolized, and DNA is also released through speaking, coughing, and the agitation of clothing. Every occupied indoor space is permeated with human genetic material, floating in the air and accumulating in dust, without anyone touching anything.

Chiara Fantinato and colleagues at the University of Oslo demonstrated this in a 2023 study published in Scientific Reports. They collected 40 air samples and 144 dust samples from offices, meeting rooms, and forensic laboratories. Dust samples were rich sources of DNA, with a median yield of 2.3 nanograms; 91 percent contained complex mixtures indicating four or more contributors. Air samples yielded less DNA but 65 percent still produced multi-person mixtures. Known occupants were successfully identified from dust samples, and non-occupants who had merely visited the space were also detected. Fantinato’s group described dust as providing a “historical record” of occupancy, capturing DNA of people present over months or years. They also noted that detectable levels of human DNA were found in air and dust from ultra-clean forensic laboratories, environments designed to be free of extraneous biological material.

In a 2024 follow-up, Fantinato and colleagues extended this line of work by analyzing samples recovered from commonly touched surfaces such as light switches and door handles in an office environment, while comparing those results to air and dust samples collected in parallel from the same areas. The study showed that surface DNA sampling can help identify occupants of a location and that combining surface, air, and dust data can improve understanding of DNA transfer in an indoor setting. For defendants, the risk is obvious. Environmental DNA may indicate that someone was present in a place but not what that person did there or when the presence occurred.

DNA in Water

A 2022 study by Marie Antony Dass and colleagues at Deakin University investigated whether human DNA could be detected in water. Using a human-specific assay targeting mitochondrial DNA, the researchers found that human eDNA remained detectable in environmental water for up to 11 days and in distilled water for up to 35 days. Partial short tandem repeat (“STR”) profiles could be recovered from environmental water for up to 24 hours. The research is preliminary, but the direction is clear. DNA is increasingly recoverable from environments, not just from objects.

What This Means for the Courtroom

A defendant’s DNA in a crime-scene dust sample does not mean the defendant committed the crime. At most, it may indicate that DNA attributable to the defendant became associated with that indoor environment, which may reflect past presence, ordinary occupancy, visiting, or indirect redistribution. The dust data do not by themselves establish what the person did there or when the DNA was deposited. A defendant’s DNA on the victim’s dog likewise does not mean the defendant assaulted the victim. It may reflect petting the animal, contact through an intermediary, or another transfer route. Each of these alternative explanations is supported by the empirical data. However, none is likely to be presented to a jury unless the defense has access to counsel who understands the science and the resources to retain an expert who can explain it.

When the Lab Is the Crime Scene

The previous sections documented how DNA travels through handshakes, household pets, air, dust, and evidence packaging, all before it reaches the laboratory. This section turns to what happens once it arrives and to the uncomfortable reality that the laboratory itself can be a source of the contamination it is supposed to detect.

Forensic DNA laboratories employ physical barriers between processing areas, mandate protective clothing, require regular decontamination, and use negative controls to flag the introduction of extraneous DNA. These protocols exist because the sensitivity of modern technology makes contamination not merely possible but predictable. When instruments can generate a profile from a handful of cells, the stray skin particle from a technician’s wrist or the residual DNA on a reused plastic tray can alter the result. And when that alteration goes undetected, the consequences affect people who have no idea the laboratory exists.

The Phantom of Heilbronn

For 16 years, police in Germany, Austria, and France believed they were hunting one of Europe’s most prolific serial criminals. Beginning in 1993, an identical female DNA profile appeared at more than 40 crime scenes, linking offenses ranging from petty thefts to six murders, including the 2007 killing of a police officer in Heilbronn. Authorities offered a reward of 300,000 euros. The press dubbed the unknown woman the “Phantom of Heilbronn.”

In March 2009, the investigation collapsed. Analysts found the Phantom’s DNA on evidence related to the burned remains of a male asylum seeker in France, an impossible result given the victim’s sex. Investigators finally reconsidered the premise they had accepted for more than a decade.

The source was a worker at an Austrian factory connected to the production or packaging of the cotton swabs used by police forensic teams across multiple jurisdictions. The swabs had been sterilized to kill bacteria, viruses, and fungi, but sterilization is not the same as DNA-free certification. DNA already present on the swabs was then carried into casework across the continent. The larger lesson was unsettlingly simple. Investigators had been chasing contamination introduced by their own collection tools rather than a real serial offender.

More than 40 crimes had been linked to the Phantom. The case prompted the International Organization for Standardization to publish ISO 18385 in 2016, the world’s first international standard on the manufacture of forensic consumables, designed to minimize the risk of human DNA contamination in products used to collect, store, and analyze biological material for forensic purposes.

Adam Scott and the Reused Tray

If the Phantom demonstrates the vulnerability of the supply chain, Adam Scott’s case demonstrates the vulnerability of the laboratory workflow. In October 2011, Scott, from Exeter, was charged with the rape of a woman in Manchester. The sole evidence was a partial DNA profile developed by LGC Forensics, which was reported at the time as carrying odds of roughly one in one billion against an unrelated match. Scott insisted he had never been to Manchester. Later cell-site analysis placed the phone he was using in Plymouth a few hours after the reported rape. He remained in jail.

LGC Forensics had recently installed robotic systems to automate DNA extraction. A technician failed to discard a disposable plastic tray after use, and the tray was reused for a subsequent batch. Scott’s DNA, legitimately present in an unrelated sample from a spitting incident processed in the same laboratory, contaminated the rape victim’s samples. The U.K. Forensic Science Regulator concluded that Scott was the “innocent victim of avoidable contamination,” citing the technician’s failure to follow disposal procedures and LGC Forensics’ failure to consider contamination despite concerns raised by the investigating officer. Scott spent approximately five months in custody before the case was withdrawn.

The regulator concluded that the tray-reuse error had occurred on at least two occasions and that roughly 26,000 samples processed during the relevant period were checked with no further identified contamination cases. The rate of detected contamination was therefore very low. But for Adam Scott, the rarity of the error did nothing to reduce its consequences.

The Systemic Dimension

These cases are among the most widely cited in the contamination literature, but they are not outliers. They sit at the visible end of a spectrum that extends into quieter failures such as mislabeled samples, cross-contamination between exhibits processed in sequence, and negative controls showing anomalous results that are explained away rather than investigated.

A 2024 paper by Kristy Martire and colleagues in Forensic Science International: Synergy warned that errors in forensic science are inevitable, not because analysts are incompetent but because the work involves complex, high-volume processes performed under time pressure. The authors argued that the forensic community’s relationship with error is often counterproductive. Scientists who acknowledge the possibility of error in court risk having their professionalism questioned, creating incentives to minimize the very vulnerabilities that quality assurance systems are designed to address. SWGDAM’s 2017 Contamination Prevention and Detection Guidelines specify requirements for physical separation, protective equipment, cleaning protocols, and negative controls. They represent best practices. But the history of forensic contamination is largely a history of implementation failures, i.e., procedures adequate on paper that were not followed and institutional cultures that treated contamination as theoretical rather than operational.

For the defendant, the practical challenge is formidable. Contamination occurs outside the defendant’s knowledge or control. Detecting it may require access to laboratory records, batch processing logs, and chain-of-custody documentation that may not be readily available to the defense in time to identify the problem. If the contaminated sample has been consumed during testing, retesting may be impossible. And even when contamination is suspected, the practical burden of uncovering it, obtaining records, retaining an expert, and presenting a coherent contamination theory often falls on the defense, which must overcome the deeply ingrained assumption that a DNA match is unquestionably correct.

The Phantom of Heilbronn and Adam Scott are instructive precisely because the contamination was eventually discovered. In one case, it took 16 years and an impossible result. In the other, it took an alert officer who questioned the DNA evidence despite its statistical weight. The harder, and more troubling, question is how many contamination events go undetected because the result is plausible, the defendant lacks resources, and no one thinks to look.

The Black Box: When Transfer Evidence Meets Probabilistic Software

The previous sections documented how DNA moves through handshakes, clothing, household pets, air, dust, packaging, and laboratory workflows, arriving on objects and at locations the identified person may never have touched or visited. These transfer mechanisms produce a distinctive category of forensic sample that is low-template, often degraded, and frequently contains genetic material from multiple contributors whose deposits overlap in ways that resist clean separation. This is precisely the category of sample that crime laboratories increasingly process through probabilistic genotyping software. And the software, by design, does not ask the question the science of transfer has shown to be essential: how did the DNA get there?

Probabilistic genotyping programs evaluate whether a person of interest is among the contributors to a mixed DNA sample. They replaced earlier methods, including a statistical framework called the combined probability of inclusion, which the 2016 report by the President’s Council of Advisors on Science and Technology (“PCAST”), found problematic for complex mixtures. The software uses mathematical modeling to produce a likelihood ratio (“LR”), a number expressing how much more probable the observed DNA results are under one proposition than another. Two systems dominate American probabilistic-genotyping litigation: (1) STRmix, which is widely used in public crime laboratories and (2) TrueAllele, offered by the Pittsburgh company Cybergenetics. Both represent a genuine advance over the more subjective and less informative approaches they displaced. But PCAST cautioned that foundational validity had been established only for limited scenarios involving samples with up to three contributors, where the minor contributor constituted at least 20 percent of the intact DNA and the tested amount exceeded the method’s minimum threshold.

The samples generated by the transfer mechanisms documented in this article routinely fall outside those parameters. A knife handled by one person after a handshake with another may carry a mixture in which the secondary contributor’s DNA is present at trace levels, well below 20 percent. A gun swabbed after passing through multiple hands, a police vehicle, and an evidence bag may yield a four- or five-person mixture with degraded, overlapping profiles. These are the samples on which probabilistic genotyping is increasingly deployed, and they are the samples for which its reliability is least established.

False Precision from Ambiguous Evidence

In December 2024, the National Institute of Standards and Technology (“NIST”) published its scientific foundation review of DNA mixture interpretation, drawing on a large body of literature and highlighting both the complexity of the field and the limits of current performance data. The report did not conclude simply that publicly available data were too sparse to permit an independent assessment of reliability. More specifically, it explained that assigning a general degree of reliability to likelihood-ratio values in global forensic casework was not feasible because there is no “true LR” for a particular casework sample, while also urging broader public release of validation, proficiency-testing, and related performance data to permit independent assessment of mixture-interpretation performance. NIJ-funded researchers at the Defense Forensic Science Center separately found significant intra- and inter-laboratory variation in mixture interpretation and reported that the majority of participating laboratories had difficulty interpreting three-person mixtures.

The consequences of this uncertainty become vivid when two programs analyze the same ambiguous sample and reach opposite conclusions. In a 2023 case study published in the Journal of Forensic Sciences, William C. Thompson documented a federal case in which STRmix and TrueAllele each assessed the same low-template mixture from a plastic bag. STRmix returned a likelihood ratio of 24 in favor of a non-contributor hypothesis. TrueAllele returned likelihood ratios ranging from 1.2 million to 16.7 million in favor of a contributor hypothesis, depending on the reference population. In plain terms, one analysis modestly favored the proposition that the person was not a contributor, while the other strongly favored the proposition that he was. Neither number measured guilt or answered how the DNA arrived on the plastic bag. The discrepancy arose from differences in analytical thresholds and statistical models and in how each program handled the low-quality peaks characteristic of trace DNA, the very kind of sample that transfer, persistence, and contamination produce.

The Code Defendants Cannot See

The problem is compounded by secrecy. In State v. Pickett, 246 A.3d 279 (N.J. Super. Ct. App. Div. 2021), New Jersey’s Appellate Division did not announce a blanket right to source-code disclosure in every probabilistic-genotyping case. Rather, it held that when the State relies on novel probabilistic genotyping software at a Frye hearing, the defense may obtain the source code and related testing, design, bug-reporting, change-log, and program-requirement materials under a protective order upon a showing of particularized need. On the facts before it, Pickett satisfied that burden.

In contrast, in United States v. Anderson, 673 F. Supp. 3d 671 (M.D. Pa. 2023), aff’d, 171 F.4th 232 (3d Cir. 2026), the federal courts held that TrueAllele was sufficiently testable and reliable for Rule 702 purposes without source-code disclosure. The Third Circuit observed that the Government had identified 42 validation studies, cited a 2014 study reporting a 0.005 percent false-positive rate, and concluded that alleged flaws in TrueAllele’s methodology or results are to be explored through cross-examination at trial, not by treating a Daubert hearing as a vehicle for compelled source-code access.

For a defendant whose DNA arrived on a weapon through secondary transfer, through a shared surface, or through packaging contamination, this combination can be devastating. The software converts ambiguous trace evidence into a number that sounds conclusive. The jury has no framework for understanding that the number addresses contributor or source-level propositions, not activity, timing, or mechanism of transfer. And depending on the jurisdiction, procedural posture, and discovery standard, the defense may be unable to inspect the proprietary code that produced the number, even when it can challenge validation studies, assumptions, input data, and interpretation through discovery, expert testimony, and cross-examination. The uncertainty that the preceding sections have documented does not disappear when it enters the algorithm. It is absorbed, processed, and returned to the courtroom masquerading as mathematical precision.

The Reports That Should Have Changed Everything

The scientific problems documented in the preceding sections did not escape the notice of the country’s most credible scientific institutions. Beginning in 2009, a succession of major national and federal reports identified these vulnerabilities, warned that the forensic system was outrunning its scientific foundations, and recommended structural reforms. The system’s response, in nearly every instance, was to resist, dilute, or dismantle the reform effort itself.

A Landmark and Its Aftermath

In February 2009, the National Research Council of the National Academy of Sciences published Strengthening Forensic Science in the United States: A Path Forward, the product of a congressionally mandated two-year investigation. The committee, which included federal judges, forensic practitioners, academic scientists, and statisticians, delivered an assessment that was blunt even by the Academy’s standards. The report concluded that the forensic science system had “serious problems” requiring a national commitment to structural overhaul. With the notable exception of nuclear DNA analysis from single-source samples, no forensic method had been rigorously demonstrated to connect evidence to a specific individual with a high degree of certainty. The report made 13 recommendations, headlined by a call for Congress to create an independent National Institute of Forensic Science to set enforceable standards, fund research, and separate the oversight of forensic disciplines from the law enforcement agencies that relied on them.

Congress never created that institute. What emerged instead was a more modest substitute in the form of the National Commission on Forensic Science, a 30-member advisory panel jointly administered by the Department of Justice and the NIST beginning in 2013. The commission included prosecutors, defense attorneys, judges, forensic practitioners, and independent research scientists. Over four years, it produced 43 documents, recommended universal accreditation for forensic laboratories, and urged practitioners to stop using the phrase “reasonable scientific certainty,” language that implied a standard of confidence that forensic methods often could not support. The commission had no enforcement power, but it represented the first sustained national dialogue between the scientific community and the legal system about how forensic evidence should be generated and presented.

The PCAST Report and the Rejection That Followed

In September 2016, the President’s Council of Advisors on Science and Technology published Forensic Science in Criminal Courts: Ensuring Scientific Validity of Feature-Comparison Methods, the most rigorous federal evaluation of forensic reliability since the NAS report. PCAST reviewed more than 2,000 scientific papers and consulted a senior advisory panel that included nine current or former federal judges and a former Solicitor General. Its conclusions were precise and unwelcome. For DNA analysis of complex mixtures, the trace samples generated by the transfer mechanisms this article has documented, PCAST found that foundational validity had been established only for limited scenarios involving no more than three contributors where the minor contributor’s DNA constituted at least 20 percent of the mixture. The council recommended that the Attorney General direct Department of Justice (“DOJ”) attorneys to ensure expert testimony met standards of scientific validity and that federal judges consider the report’s criteria when ruling on admissibility.

The institutional response was swift and negative. Attorney General Loretta Lynch stated that the DOJ would not adopt the report’s recommendations, expressing confidence that existing legal standards were “based on sound science and sound legal reasoning.” The FBI published its own critique, accusing PCAST of using “subjectively derived” criteria and ignoring relevant validation studies. The National District Attorneys Association warned that adopting any of the recommendations would have “a devastating effect” on the ability of law enforcement and prosecutors to investigate and try cases. When PCAST asked the DOJ to identify the specific studies it claimed had been omitted, the department conceded by December 2016 that it could identify none.

The structural dismantling followed. In April 2017, Attorney General Jeff Sessions announced that the DOJ would not renew the National Commission on Forensic Science. The independent panel that had brought research scientists and criminal justice stakeholders to the same table was replaced by an internal DOJ working group led by a career prosecutor who had been one of only two commission members to vote against standardizing the language forensic examiners use in testimony. The department simultaneously suspended an ongoing review of FBI forensic testimony that had been launched after the FBI admitted in 2015 that examiners in one of its laboratories had given flawed testimony in hundreds of cases.

The Reports That Kept Coming

The dissolution of the commission did not halt the science. In 2018, NIST published the results of two interlaboratory studies, designated MIX05 and MIX13, in which 69 and 108 accredited crime laboratories, respectively, were given identical DNA mixture data and asked to interpret them. The results revealed striking variation. In the MIX13 study’s most complex scenario, a four-person mixture from a mock ski mask found near a bank robbery scene, 74 out of 108 participating laboratories falsely included a known non-contributor using the combined probability of inclusion method that many labs still employed. The study’s lead author, NIST geneticist John Butler, told Forensic Magazine that the scenario was designed specifically to test whether the method could falsely place someone at a crime scene. “It demonstrated that really nicely,” Butler said. The MIX13 data had been collected in 2013. The peer-reviewed paper was not published until August 2018, five years later. Critics, including Boise State University geneticist Greg Hampikian, argued that earlier publication could have provided criminal defendants with admissible scientific evidence to challenge unreliable mixture interpretations during the intervening years.

In May 2024, NIST and NIJ published the most comprehensive assessment yet: Forensic DNA Interpretation and Human Factors: Improving Practice Through a Systems Approach, the product of an expert working group that convened in 2020 and included forensic scientists, cognitive psychologists, statisticians, and legal scholars. The report issued 44 recommendations spanning cognitive bias, DNA transfer and persistence, quality assurance, testimony language, and laboratory culture. It explicitly addressed the transfer science at the core of this article, recommending that forensic service providers develop protocols for considering how DNA may have been deposited before rendering conclusions. The report acknowledged what the preceding sections have documented from the published science, namely that DNA presence alone does not establish the circumstances of its deposit and that human judgment at every stage of the analytical process introduces the potential for error.

The pattern across 17 years is consistent. The most credible scientific bodies in the country have examined the forensic system and reached variations of the same conclusion. The gap between what the science can support and what the courtroom assumes it proves is dangerous, and the institutional structures meant to close that gap are either absent, voluntary, or actively opposed by the agencies that benefit most from the status quo. The NAS called for independent oversight. It was never created. The commission that substituted for it was disbanded. PCAST’s recommendations were rejected before the ink dried. And the laboratories whose work these reports scrutinized continued interpreting the trace, degraded, multi-contributor samples produced by DNA transfer, samples whose ambiguity the preceding sections have shown to be inherent, under rules and assumptions that the country’s leading scientists had told them, repeatedly, were insufficient.

The Jury Problem: CSI Effect, Misleading Terminology, and Prosecutorial Overclaiming

The reports discussed in the previous section were directed at scientists, policymakers, and judges. They addressed laboratory protocols, validation studies, and institutional architecture. But the forensic system’s most consequential audience is none of those. It is the people in the jury box, and the problems that arise when DNA evidence reaches them are different in kind from those that arise in the laboratory. Jurors do not misinterpret DNA evidence because they are unintelligent. They misinterpret it because the language in which it is presented, the assumptions embedded in that language, and the arguments prosecutors build around it systematically invite the exact inferential leap the science cannot support, i.e., from presence to guilt.

What Jurors Believe About DNA

Research on juror attitudes toward forensic evidence has long shown that scientific evidence carries unusual force in the courtroom. A 2008 NIJ-funded survey of 1,027 prospective jurors in Ann Arbor, Michigan, found that 46 percent expected to see some form of scientific evidence in every criminal case and that expectations for DNA evidence rose with offense type, including 46 percent in murder or attempted-murder cases and 73 percent in rape cases. However, that study found only scant and inconsistent evidence that CSI viewers were systematically more likely to acquit when scientific evidence was absent. A separate 2017 mock-juror study by forensic psychology researcher Logan Ewanation and colleagues found that jurors were more likely to vote guilty when presented with DNA or fingerprint evidence than with eyewitness testimony and that evidence strength affected verdicts only in the eyewitness condition. Taken together, the studies support a narrower but important point. Jurors often attach exceptional persuasive weight to forensic evidence, especially DNA.

Whether this constitutes a true “CSI effect” remains debated among researchers, and the NIJ study itself found only inconsistent differences between CSI viewers and nonviewers on acquittal rates. But the more consequential finding, one that cuts across the debate, is that DNA evidence carries an aura of infallibility in the jury room that other forms of evidence do not. Jurors treat it less as one piece of a puzzle and more as a definitive answer. And for the categories of DNA evidence this article has examined, that confidence is misplaced.

The Words That Do the Damage

The problem begins with terminology. As explained earlier, the phrase “touch DNA” encodes a factual assumption the science cannot verify. That is, the identified person physically touched the object on which the DNA was found. The term is so embedded in forensic practice that analysts, prosecutors, and even judges use it reflexively, often without recognizing its prejudicial freight. When a jury hears that “the defendant’s touch DNA was recovered from the weapon,” the phrase does much of the prosecution’s inferential work before any argument has been made.

Courts have begun to take notice. In State v. Phillips, 844 S.E.2d 651 (S.C. 2020), the South Carolina Supreme Court reversed a murder conviction after the State introduced touch-DNA mixture evidence from both the victim’s gun and the inside of his jeans pocket. The analyst testified that Phillips could not be excluded as a contributor but also acknowledged that the probability an unrelated person could not be excluded was about 1 in 200 for the gun sample and 1 in 2 for the pocket sample. The Court faulted the State for failing to explain the underlying concepts of touch DNA, non-exclusion DNA, and random-match probability, and it pointed specifically to the prosecutor’s closing-argument misstatements, including that if Phillips had not touched the gun or pocket, his DNA would not be there. Because no meaningful Daubert / Council hearing had been conducted, the Court held that the State failed to establish the Rule 702 “assist the trier of fact” element and that the evidence’s probative value was substantially outweighed by the danger of confusing and misleading the jury.

The Phillips decision was unusual in its willingness to engage the science directly. A more recent proceeding underscored how difficult the terminology problem remains. In the Idaho prosecution of Bryan Kohberger for the 2022 University of Idaho student murders, the defense moved to exclude the terms “touch DNA” and “contact DNA,” arguing that both phrases assume a fact the prosecution must prove – that the defendant physically handled the knife sheath on which his DNA was found. In its April 18, 2025 order, the Court declined to prohibit the term “touch DNA” and was not persuaded that the jury would be misled, though it acknowledged that the term can be misleading, asked counsel to avoid the disputed terms as much as possible, and directed the parties to submit a mutually agreeable instruction for use if the terms were inadvertently used. The same order also concluded that the State’s rebuttal transfer opinions were proper. That is a meaningful distinction. The Court did not bar the terminology outright, but it also did not treat the concern as frivolous.

When Language Becomes Argument

The terminology issue is not limited to “touch.” Forensic reports and testimony sometimes use phrases like “the wearer of the glove,” “the user of the weapon,” or “the handler of the item” to describe the person whose DNA was found, language that can encourage jurors to treat source-level evidence as proof of activity. NIST’s 2024 human-factors report addresses the same underlying problem more carefully. It warns that profile descriptors should not be interpreted as telling the factfinder about “the mechanisms, actions, or timing” that led to DNA deposition, and it explains that a DNA comparison result is not meaningful for answering how or when DNA was deposited. That is the more defensible way to frame the risk.

The deeper structural problem is that the courtroom framework gives prosecutors wide latitude to build narratives around DNA evidence that the science does not support. The likelihood ratio produced by probabilistic genotyping software tells the jury how much more probable the observed genetic data are under a specified proposition that the defendant contributed to the mixture than under a specified alternative contributor proposition. It does not tell the jury how probable it is that the defendant committed the charged act. That distinction is the most important conceptual boundary in forensic DNA interpretation, and it is the boundary most consistently erased in practice. A prosecutor who tells a jury that the defendant’s DNA was found on the victim’s clothing and then argues that this proves the defendant attacked the victim has made an activity-level claim on the basis of source-level evidence. The jurors, who have absorbed decades of cultural messaging about the certainty of DNA, have no independent framework for recognizing the gap.

This is not a hypothetical concern. The wrongful conviction literature, which the next section will examine, is populated by cases in which DNA evidence of marginal probative value was presented to juries in language that maximized its apparent significance. The science documented earlier in this article creates a universe of plausible alternative explanations for DNA presence. But those explanations reach the jury only if the defense has the resources and expertise to present them, and the judge has the scientific literacy to evaluate them. The reports discussed above were written, in part, to close precisely this knowledge gap. The system’s resistance to those reports means the gap persists, and it persists most acutely in the room where it matters most.

The Human Toll: Lydell Grant and the Wrongful Conviction Machine

On the night of December 10, 2010, 28-year-old Aaron Scheerhoorn was stabbed to death outside Club Blur, a bar in Houston’s Montrose neighborhood. The following evening, Lydell Grant, a 33-year-old Black man, parked his car in the lot shared by the bar and an adjacent establishment. An employee of Club Blur who had witnessed the murder the night before saw Grant and thought he resembled the attacker. The employee wrote down Grant’s vehicle identification number and called Crime Stoppers.

Based on that tip, Houston police assembled a photo lineup of six Black men that included Grant. Over the next day and a half, witnesses identified Grant from the spread. The lead detective administered the identification process in a non-double-blind manner, meaning he knew which photograph was the suspect as he displayed the spread and observed the witnesses’ reactions. Six months later, the Texas Legislature would pass a law discouraging that kind of identification procedure. But by then, Grant had been arrested, charged with murder, and was awaiting trial.

At trial in December 2012, the prosecution’s case rested overwhelmingly on the eyewitness identifications. But jurors also heard from a forensic DNA analyst with the Houston Police Department Crime Laboratory, who testified about results from fingernail scrapings taken from Scheerhoorn’s right hand during autopsy. The scrapings contained a mixture of the victim’s DNA and that of at least one unknown male. Hill initially testified that no conclusion could be made regarding Grant as a contributor to that mixture. On further questioning by the prosecutor, she agreed that Grant could not be excluded as a potential contributor. The jury heard the eyewitnesses say Grant was the killer and heard the DNA analyst say he could not be ruled out. Grant had an alibi witness who placed him elsewhere that night. The jury convicted him, and he was sentenced to life.

What the DNA Actually Showed

Grant contacted the Innocence Project of Texas from prison. Mike Ware, the organization’s executive director, was skeptical of the eyewitness testimony but particularly troubled by the DNA. The Houston crime lab analyst had been unable to deconvolute the mixture at the time of trial. Ware arranged for the fingernail evidence to be retested using TrueAllele, the probabilistic genotyping software discussed earlier. The reanalysis separated the mixture and generated a profile for the unknown male contributor. That profile was uploaded to the FBI’s Combined DNA Index System (“CODIS”).

The CODIS search returned a hit: Jermarico Carter, a Houston man with a violent criminal history who fit the general description witnesses had given of the attacker, though available accounts indicate that Carter and Grant did not closely resemble one another beyond both being Black men. Carter had previously been arrested on a drug charge within feet of where the murder occurred, left Houston shortly after the killing, and relocated to Atlanta. When Houston police later interviewed Carter after his arrest in Georgia, he confessed to stabbing Scheerhoorn.

In May 2021, the Texas Court of Criminal Appeals granted Grant actual-innocence relief. The Court concluded that the record supported the trial court’s determination that Grant had established actual innocence by clear and convincing evidence, set aside the murder judgment, and remanded him to Harris County custody to answer the indictment. By then, Grant had served nearly nine years of a life sentence for a murder committed by someone else.

The Mechanism of Failure

Grant’s case is a textbook study in the interaction between the scientific limitations this article has documented and the systemic pressures of criminal prosecution. The DNA on Scheerhoorn’s fingernails was a transfer deposit, biological material consistent with the kind of skin-cell evidence discussed in previous sections, recovered from a surface where direct physical contact during a violent struggle would be expected. The presence of unknown male DNA under a stabbing victim’s nails is, in that context, genuinely probative. But the Houston crime lab could not determine whose DNA it was. The analyst’s testimony that Grant “could not be excluded” communicated almost nothing to the jury while sounding like something of considerable value. It did not mean Grant’s DNA was there. It meant the data were too ambiguous to say either way. Combined with six confident eyewitnesses, that ambiguity was enough to convict.

The DNA evidence in Grant’s case failed at the interpretive level, not the collection level. The mixture existed. The question was whether existing technology could resolve it and whether the testimony accurately conveyed what the analysis had, and had not, accomplished. As a detailed case study in The ISHI Report noted, the original analyst’s characterization of the results was misleading. Postconviction testing excluded Grant, confirmed an unknown male contributor, and produced a profile that could be searched in CODIS. Mike Ware later described Grant’s case as historic because it showed how newer DNA analysis could reopen wrongful-conviction cases that once appeared closed. That is the safer and better-supported way to describe the significance of the DNA findings here.

A Pattern, Not an Anomaly

Grant’s case illustrates dynamics that repeat across the wrongful conviction landscape. As of 2023, the National Registry of Exonerations (“NRE”) had documented more than 3,000 wrongful convictions in the U.S. The Innocence Project’s DNA-exoneration data page reports 375 DNA exonerations nationwide through 2020, including 21 people who had served time on death row. The same page states that the Innocence Project worked on 190 of those DNA exonerations.

A 2023 NIJ-funded study by John S. Morgan examined 732 of those cases, those classified by the NRE as involving false or misleading forensic evidence, analyzing 1,391 individual forensic examinations. Morgan’s findings challenged the “bad apple” narrative that attributes forensic failures to rogue analysts. Most errors were not identification or classification mistakes by laboratory technicians. They were testimony and communication failures. That is, experts who mischaracterized statistical weight, reports that omitted critical limiting information, and prosecutors who framed ambiguous results as definitive. Actors outside the forensic laboratory – investigators who suppressed evidence, prosecutors who misrepresented results, defense attorneys who lacked the resources to mount effective challenges – contributed substantially to the errors. Morgan concluded that in roughly half the wrongful convictions he examined, improved technology, testimony standards, or practice standards could have prevented the conviction at the time of trial.

That finding circles back to every section of this article. The science of DNA transfer, persistence, and contamination documented in earlier sections creates a universe of ambiguity that the forensic system is not structured to communicate honestly. The reports discussed identified the problem and proposed solutions. The courtroom dynamics described show how that ambiguity is erased by terminology, narrative, and juror predisposition. And the wrongful conviction data confirm that these structural failures result in real people, serving real sentences, for crimes they did not commit, on the basis of forensic evidence whose limitations were never explained to the people deciding their fate. Grant lost nearly a decade. The system that convicted him did not malfunction in any single dramatic way. It worked exactly as designed, and the design was inadequate to the complexity of the evidence it was processing.

DNA on Death Row: Marcellus Williams, Rodney Reed, and the Fight for Testing

The wrongful conviction of Lydell Grant cost him nearly a decade. On death row, the stakes collapse into a single question. Will the state execute a person whose guilt the available DNA evidence does not confirm, and in some cases actively contradicts, because the legal system lacks the will or the mechanism to test what should be tested before it is too late?

Marcellus Williams and the Contaminated Knife

Marcellus Williams, a 55-year-old Black man, was executed by the state of Missouri on September 24, 2024. He had spent 23 years on death row for the 1998 stabbing death of Felicia Gayle, a former newspaper reporter, in her St. Louis–area home. No eyewitness, fingerprint, or forensic evidence placed Williams at the scene. The prosecution’s case rested primarily on the testimony of two informants, a former girlfriend and a jailhouse witness, both of whom were incentivized by reward money and the prospect of leniency in their own criminal matters.

In 2016, DNA testing of the handle of the butcher knife used to stab Gayle 43 times revealed male DNA that did not match Williams. Three independent DNA experts concluded he was not the source. A board of inquiry appointed by then-Governor Eric Greitens in 2017 to investigate the DNA evidence was dissolved by his successor, Mike Parson, before completing its work. When St. Louis County Prosecuting Attorney Wesley Bell moved in January 2024 to vacate Williams’ conviction based on the DNA and other evidence, subsequent testing revealed that the knife had been handled without gloves by the original trial prosecutor and an investigator, thereby contaminating the very evidence that might have identified the actual killer. The State Attorney General opposed vacatur. The Missouri Supreme Court allowed the execution to proceed. On September 24, 2024, the U.S. Supreme Court denied an application for a stay of execution and denied certiorari. The Court’s order noted that Justices Sotomayor, Kagan, and Jackson would have granted the stay. Missouri executed Williams later that day.

Williams’ case is a direct illustration of the contamination dynamics described in this article. The murder weapon sat in evidence for years, handled by ungloved officials, accumulating DNA deposits that had nothing to do with the crime. By the time modern testing could have resolved the question, the evidence was irrevocably compromised, not by the defendant but by the system that convicted him.

Rodney Reed and the Belt That Cannot Be Tested

Rodney Reed has been on Texas’ death row for nearly 30 years, convicted in 1998 of the 1996 murder of Stacey Lee Stites in Bastrop County. Reed has maintained that he was having a consensual affair with Stites and that her fiancé, Jimmy Fennell, a Bastrop County police officer later convicted of an unrelated sexual assault, committed the murder. Reed has proffered evidence to support this claim, including sworn affidavits attesting that Fennell told a colleague one month before the murder that Stites was involved with a Black man and that Fennell, while imprisoned in 2019 for the sexual assault conviction, confessed to a fellow prisoner that he killed Stites. The prosecution’s case turned on the presence of Reed’s semen, which he contends is explained by the consensual relationship.

For over a decade, Reed has sought DNA testing of the murder weapon, a webbed belt used to strangle Stites. As Justice Sotomayor noted in her 2026 dissent, a significant amount of the killer’s DNA is likely to be on the belt because the killer, in an act of “great force,” used it to strangle Stites for approximately three to four minutes. The belt is precisely the kind of sustained-contact surface on which the transfer deposits described throughout this article would concentrate in recoverable quantities. If the DNA on that belt belongs solely to Reed or solely to Fennell, that finding could resolve the case. However, Texas courts have denied testing on the grounds that the belt was “contaminated” after being handled by ungloved attorneys, court personnel, and possibly jurors, a chain-of-custody failure that Texas’ post-conviction DNA testing statute, Article 64, treats as a bar to testing.

Reed raised three arguments before the Fifth Circuit challenging that bar. First, he argued that the noncontamination requirement is scientifically arbitrary because modern DNA testing can produce reliable results even from contaminated evidence. A former lead forensic scientist at the Texas Department of Public Safety, Chase Baumgartner, submitted an amicus brief stating that even in a worst-case scenario of a complex, contaminated DNA profile, state analysts could accurately include or exclude Reed or Fennell with above 95 percent accuracy. Second, Reed argued that it is fundamentally unfair to hold contamination against the prisoner when the State is responsible for the condition of the evidence while it is stored and handled. Third, he noted that Texas courts routinely admit DNA test results from contaminated evidence when offered by the prosecution to secure a conviction but invoke contamination to deny testing that might exonerate a condemned man. The Fifth Circuit rejected the second and third arguments but, as Sotomayor’s dissent observed, never squarely addressed the first. That is, it never confronted the question whether blocking testing serves any legitimate purpose when the science can account for contamination and still yield accurate results.

In Gutierrez v. Saenz, 606 U.S. 305 (2025), the Supreme Court held that Gutierrez had standing to bring a § 1983 due-process challenge to Texas’ postconviction DNA-testing procedures. Justice Sotomayor’s opinion for the Court explained that a declaratory judgment that Texas’ postconviction DNA-testing procedures violate due process would redress the prisoner’s injury by eliminating the prosecutor’s reliance on Article 64 as a reason for denying testing and that the possibility a prosecutor might later find some other reason to deny access does not defeat standing.

But when the Supreme Court considered Reed’s own petition in March 2026, a majority denied certiorari, leaving the Fifth Circuit’s judgment in place. Reed v. Goertz, 607 U.S. ___ (2026) (Sotomayor, J., dissenting from denial of certiorari). Justice Sotomayor, joined by Justices Kagan and Jackson, called the State’s refusal to allow testing “inexplicable” given the “very substantial possibility that such testing could exculpate Reed and identify the real killer.” Without the Court’s intervention, she wrote, “the State will likely execute Reed without the world ever knowing whether Reed’s or Fennell’s DNA is on the murder weapon, even though a simple DNA test could reveal that information.”

The Structural Problem

These cases share a structural feature that connects them to every section of this article. The science of DNA transfer, persistence, and contamination has demonstrated that biological material moves through environments in ways that are often invisible and always uncontrollable. That same science has produced tools – probabilistic genotyping, advanced STR analysis, touch DNA recovery methods – capable of resolving questions that were unanswerable at the time of trial. However, the legal system has not built procedural architecture adequate to the science it depends on. States create postconviction testing statutes and then narrow them with requirements such as contamination exclusions, actual-innocence thresholds, and timeliness bars that function in practice to prevent the very testing the statutes nominally authorize.

The Death Penalty Information Center has documented the pattern across jurisdictions, observing that the limitations of DNA evidence in innocence cases are compounded by procedural barriers that make it exceedingly difficult for condemned prisoners to access testing even when the technology exists to resolve the question. The result is a system in which the state can simultaneously rely on DNA evidence to convict and refuse to allow DNA testing to exonerate, deploying the science’s authority when it serves the prosecution and withholding its capacity when it might serve the defense. When the consequence of that asymmetry is death, the failure is not merely procedural. It is irreversible.

The Next Frontier of Risk: Investigative Genetic Genealogy and the Expansion of Genetic Surveillance

In April 2018, California investigators announced that, using a technique most law enforcement agencies had never heard of, they had identified Joseph James DeAngelo as the Golden State Killer, responsible for at least 13 murders and 45 rapes across the state in the 1970s and 1980s. Investigators had uploaded a crime-scene DNA profile to GEDmatch, a public genetic genealogy database populated by consumers exploring their ancestry, and used the resulting list of partial genetic matches to build a family tree that eventually led to DeAngelo. The case was a sensation. According to Baylor College of Medicine researchers, investigative genetic genealogy (“IGG”) has since been used in connection with hundreds of cases in the U.S. involving criminal perpetrators and unidentified human remains. It has also created an entirely new category of forensic risk, one that magnifies every vulnerability documented in the preceding sections of this article.

IGG works differently from the standard DNA database search that most people associate with forensic identification. CODIS compares STR profiles derived from crime-scene evidence against a database of profiles collected from convicted offenders and, in some jurisdictions, arrestees. A CODIS search either produces a direct match or it does not. In contrast, IGG converts a crime-scene sample into a single nucleotide polymorphism (“SNP”) profile, a far denser genetic map consisting of hundreds of thousands of markers, and uploads it to consumer genealogy databases such as GEDmatch or FamilyTreeDNA. Typically, the database returns not a direct match but a list of the offender’s genetic relatives such as second cousins, third cousins, and sometimes more distant kin who happen to have submitted their own DNA for ancestry research. In the unusual case where the offender is himself in the database, an exact SNP match is possible. Investigators and forensic genealogists then combine the kinship data with publicly accessible genealogical records to construct a family tree and identify high-likelihood suspects. Once a suspect is identified, law enforcement collects a DNA sample, often surreptitiously from a discarded item, and performs conventional STR comparison to confirm or exclude.

The technique is powerful. It is also, for purposes of this article, dangerous in a specific and underappreciated way. Every IGG investigation begins with a crime-scene sample. If that sample is the product of secondary or tertiary transfer, then the genealogical search, the family tree, the targeted testing, and the eventual suspect identification can all flow from a false premise. The investigator is not tracing the killer’s relatives. He is tracing the relatives of a person whose DNA arrived at the scene through a handshake, a shared surface, a pet, an ambulance, or a contaminated evidence package. The case of Lukis Anderson, discussed at the beginning of this article, illustrates the architecture of the danger. But it should be framed as a conditional risk, not a certainty in that specific case, because not every forensic sample is suitable for generating the high-density SNP profile required for IGG.

Michael Usry’s experience previewed this danger. In 2014, Usry, a New Orleans filmmaker, became a suspect in a cold-case murder investigation after his father’s participation in a now-defunct Y-chromosome genealogy database led law enforcement to consider him. Based on that genetic lead and circumstantial factors, including Usry’s possible access to the crime scene and his production of a horror film, investigators obtained a warrant to collect and test his DNA. The results excluded him, but nearly a month passed before they were returned. Usry was never arrested, but the episode demonstrated that an innocent person could be subjected to investigation, surveillance, and the collection of intimate biological material because of a genetic association he did not create and could not control.

As medical ethicist Christi Guerrini and colleagues observed in 2021, the stress of such an investigation can be significant even when the system ultimately reaches the correct result. The DOJ’s 2019 Interim Policy requires confirmatory direct STR comparison before arrest, but that policy’s reach extends only to investigations within its defined scope – including DOJ components, investigations using DOJ funding, and matters in which DOJ personnel or contractors conduct the genealogy step – rather than to every state or local IGG investigation nationwide.

That policy gap is itself a source of concern. The DOJ Interim Policy limits forensic genetic genealogical DNA analysis and searching to unsolved violent crimes and the unidentified remains of suspected homicide victims, prohibits arrests based solely on a genetic genealogy database association, and bars law enforcement from using samples or information obtained during forensic genetic genealogy to determine a person’s genetic predisposition for disease, medical condition, or psychological trait. But the policy does not have the force of law and expressly disclaims creating enforceable rights. It does not bind purely state or local investigations outside its scope. And although it does impose destruction requirements for third-party reference samples, derivative forensic genetic genealogy profiles, and genealogy-service data, broader questions about uniform governance of private laboratories, genealogists, and state-level retention practices remain unresolved. A 2025 Policy Delphi study led by Guerrini and colleagues identified nine IGG practices warranting priority policy attention, with especially strong concern about law-enforcement participation in databases against their terms of service and about management of data and samples collected or generated during IGG. State regulation remains a patchwork.

The implications are direct. IGG expands the universe of people who can be drawn into a criminal investigation through no action of their own. A person who has never submitted a DNA sample to any database can become a suspect because a third cousin participated in a genealogy service for recreational purposes. The genetic information in those databases, SNP profiles containing hundreds of thousands of markers, is orders of magnitude more revealing than the 20 STR loci stored in CODIS, carrying information about ancestry, appearance, and medical predisposition. And because IGG investigations begin with a crime-scene sample, every limitation of that sample – transfer, contamination, mixture, degradation – propagates silently through the genealogical search. The technique does not create new evidence. It creates a new way to follow old evidence, and if the old evidence is wrong, the new pathway leads to the wrong family.

What Defense Lawyers, Defendants, and Judges Should Ask and What Must Change

The evidence assembled in the preceding sections leads to a conclusion that is simple to state and difficult to act on. The criminal legal system’s treatment of DNA evidence is structurally misaligned with what the science actually shows. DNA is treated as dispositive when it is often merely associative. Presence is treated as proof of participation when the science of transfer, persistence, and contamination demonstrates that presence frequently proves nothing of the kind. The gap between what DNA can tell us and what the system assumes it tells us is not a theoretical concern. It is the mechanism by which innocent people are convicted and guilty people go free.

Closing that gap requires action on three fronts: (1) what defense lawyers must demand in individual cases, (2) what judges must understand about the evidence they admit, and (3) what institutional reforms are necessary to prevent the system from continuing to treat DNA as a magic bullet.

What Defense Lawyers Should Ask

Every case involving touch, trace, or transfer DNA should prompt a series of questions that most defense attorneys are not currently trained to ask. First, was the DNA deposited through direct contact with the item, or could it have arrived through secondary or tertiary transfer? The Cale handshake-to-knife study demonstrated that a person’s DNA can appear on a weapon they have never touched. If the prosecution’s theory requires direct contact, the defense must demand evidence supporting that inference, not simply the presence of a profile.

Second, when was the DNA deposited? DNA has no timestamp. A profile recovered from a surface may reflect contact that occurred hours, days, or weeks before or after the charged offense. Background DNA, the genetic material each person sheds continuously into the environments they inhabit, can persist on surfaces long after the person has left. If the prosecution cannot establish the temporal relevance of the deposit, the evidence does not support an activity-level inference.

Third, what happened to the evidence between collection and analysis? The packaging-transfer research and the contamination cases documented in previous sections demonstrate that DNA can move between items in an evidence bag, from an analyst’s hands to a sample, or from one case to another through shared laboratory equipment. Defense counsel should request full chain-of-custody documentation, laboratory contamination logs, and, where probabilistic genotyping software was used, the input data, software version, and analytical parameters that produced the likelihood ratio presented to the jury.

Fourth, in IGG cases, was the crime-scene sample on which the genealogical search was based a robust single-source profile, or a low-template mixture susceptible to interpretation variability? An IGG investigation built on a transferred or contaminated sample does not merely identify the wrong individual; it identifies the wrong family.

Finally, defense counsel should insist that the prosecution’s DNA expert address the evidence at the activity level, not merely the source or sub-source level. The crucial question is not “whose DNA is this?” but “how did this DNA get here, and does its presence support the prosecution’s account of what happened?” As the ISFG DNA Commission, SWGDAM, and the U.K. Forensic Science Regulator have each recognized, source-level opinions do not answer the activity-level question. If the expert cannot or will not address how the DNA arrived, the evidence should not be presented as though it answers that question.

What Judges Must Understand

Judges serve as gatekeepers for expert testimony, and the reliability of DNA evidence in any given case depends on whether the court understands what the evidence can and cannot prove. Two reforms are essential. First, courts should require that DNA experts who testify about the significance of a profile be prepared to address transfer, persistence, and contamination as alternative explanations, not as speculative possibilities but as documented scientific phenomena with an extensive empirical literature. The South Carolina Supreme Court’s decision in Phillips, reversing a murder conviction after the State overstated weak touch-DNA mixture evidence and failed to establish that the testimony would assist the jury, illustrates what happens when this gatekeeping fails.

Second, courts considering the admissibility of probabilistic-genotyping results should require meaningful defense access to the materials necessary to test the software’s reliability. Pickett did not announce a blanket right to source-code disclosure in every case. It held that, when the State relies on novel probabilistic-genotyping software at a Frye hearing, a defendant who demonstrates particularized need may obtain source code and related testing, design, bug-reporting, change-log, and program-requirement materials under a protective order. Anderson reflects the narrower federal approach. The Third Circuit held that TrueAllele was reliable enough for Rule 702 purposes without compelled source-code disclosure and that alleged flaws in the software’s methodology or results should be explored through cross-examination at trial.

What Must Change

Individual case advocacy, however effective, cannot remedy structural failures. Three institutional reforms would materially reduce the risk of DNA-based wrongful convictions.

First, forensic laboratories should be required to address the activity level when reporting DNA results in criminal cases. The U.K.’s Forensic Science Regulator has moved in this direction, publishing guidance in 2026 that establishes principles for interpretation and communication of forensic observations at multiple levels of issue, including the activity level, supported by a statutory Code of Practice with enforcement authority. The U.S. has no equivalent. SWGDAM’s 2025 position statement acknowledges the importance of activity-level evaluation but does not mandate it. Until reporting standards require laboratories to distinguish between “this is the defendant’s DNA” and “this DNA arrived here because the defendant committed this act,” the gap between science and testimony will persist.

Second, postconviction DNA testing statutes must be reformed to eliminate procedural barriers that can block potentially exculpatory testing. In Gutierrez, the Supreme Court held that Gutierrez had standing to pursue a § 1983 due-process challenge to Texas’ postconviction DNA-testing procedures, but it did not resolve the merits of that challenge. The substantive constitutional question therefore remains open. Reed shows what is at stake. Justice Sotomayor wrote that it was “inexplicable” for Texas to refuse testing despite the “very substantial possibility” that such testing could exculpate Reed and identify the real killer, and she warned that the State may execute Reed without anyone ever knowing whether Reed’s or Jimmy Fennell’s DNA is on the murder weapon, “even though a simple DNA test could reveal that information.”

Third, the opacity of probabilistic genotyping should end. Leading researchers have argued that courtroom use of probabilistic-genotyping software should be far more transparent, because peer review and validation studies are not substitutes for robust independent testing of software implementation. Some courts have required or permitted broader disclosure under protective orders. In Pickett, for example, the New Jersey Appellate Division required production of TrueAllele source code and related materials for a Frye hearing upon a showing of particularized need. The federal approach is narrower. In Anderson, the Third Circuit held that TrueAllele was admissible under Rule 702 without compelled source-code disclosure on the record before it, emphasizing the disclosures that had been made and the role of cross-examination. The safer conclusion is not that existing law already compels code access in every case but that meaningful transparency is often necessary to permit effective scrutiny of software-generated forensic evidence.

None of these reforms are radical. They ask only that the legal system treat DNA evidence with the rigor that the science demands – that presence not be confused with proof, that transfer be acknowledged as a documented reality rather than a speculative defense theory, and that the extraordinary power of genetic identification be matched by an equally extraordinary commitment to getting it right. The science has been clear for years. What remains is the will to act on it.  

 

Sources: Katie Worth, Framed for Murder by His Own DNA, Marshall Project/FRONTLINE (PBS) (Apr. 19, 2018); Peter Andrey Smith, When DNA Implicates the Innocent, 314 Sci. Am. 11 (June 2016); Innocence Project, DNA Exonerations in the United States (1989–2020); Roland A.H. van Oorschot & Maxwell K. Jones, DNA Fingerprints from Fingerprints, 387 Nature 767 (1997); Pallavi Kumari et al., Touch DNA in Forensic Science: A Comprehensive Overview and Current Status of the Research, 16 Egyptian J. Forensic Sci. 23 (2026); Pamela Tozzo et al., Touch DNA Sampling Methods: Efficacy Evaluation and Systematic Review, 23 Int’l J. Molecular Sci. 15541 (2022); Cynthia M. Cale et al., Could Secondary DNA Transfer Falsely Place Someone at the Scene of a Crime?, 61 J. Forensic Sci. 196 (2016); Nat’l Inst. of Just., Improving Analysis of “Trace DNA” Evidence (Jan. 16, 2024); FBI, Quality Assurance Standards for Forensic DNA Testing Laboratories (effective July 1, 2025); Peter Gill et al., DNA Commission of the International Society for Forensic Genetics: Assessing the Value of Forensic Biological Evidence — Guidelines Highlighting the Importance of Propositions. Part II: Evaluation of Biological Traces Considering Activity Level Propositions, 44 Forensic Sci. Int’l: Genetics 102186 (2020); Tacha Hicks et al., A Logical Framework for Forensic DNA Interpretation, 13 Genes 957 (2022); Sci. Working Grp. on DNA Analysis Methods, Position Statement on Formal Activity Level Reporting (Dec. 2025); Janet Stacey et al., Current State and Barriers to Global Adoption of Forensic Evaluative Reporting of Findings Given Activity-Level Propositions, 65 Sci. & Just. 101336 (2025); Forensic Sci. Regulator (UK), Forensic Science Activities: Interpretation and Communication (updated Mar. 24, 2026); Francesco Sessa et al., Indirect DNA Transfer and Forensic Implications: A Literature Review, 14 Genes 2153 (2023); Roland A.H. van Oorschot et al., DNA Transfer in Forensic Science: Recent Progress Towards Meeting Challenges, 12 Genes 1766 (2021); Duncan Taylor et al., Helping To Distinguish Primary from Secondary Transfer Events for Trace DNA, 28 Forensic Sci. Int’l: Genetics 155 (2017); Christopher Zoukis, Secondary DNA Transfer: The Rarely Discussed Phenomenon, Criminal Legal News (Aug. 15, 2018); Roland A.H. van Oorschot et al., DNA Transfer in Forensic Science: A Review, 38 Forensic Sci. Int’l: Genetics 140 (2019); Meghan Ramsey, Persistence of Touch DNA for Forensic Analysis, NIJ Grant No. 2018-DU-BX-0192, MIT Lincoln Laboratory (2021); Nat’l Inst. of Just., Persistence of Touch DNA for Analysis (June 5, 2023); Cara Woollacott et al., The Transfer, Prevalence, Persistence, and Recovery of DNA from Body Areas in Forensic Science: A Review, 5 Forensic Sci. 9 (2025); Yong Sheng Lee & Christopher Kiu-Choong Syn, DNA Transfer Between Items Within an Evidence Package, 16 Genes 894 (2025); Heidi Monkman, Bianca Szkuta & Roland A.H. van Oorschot, Presence of Human DNA on Household Dogs and Its Bi-Directional Transfer, 14 Genes 1486 (2023); Heidi Monkman, Roland A.H. van Oorschot & Mariya Goray, The Role of Cats in Human DNA Transfer, 74 Forensic Sci. Int’l: Genetics 103132 (2025); Heidi Monkman, Roland A.H. van Oorschot, Volgin Luke & Mariya Goray, Investigation of Human DNA Transfer During Mock Dog-Napping, 378 Forensic Sci. Int’l 112724 (2026); Carla Bini et al., A Mini Narrative Review on Human DNA Transfer Involving Dogs and Cats and Their Role in Forensic Investigation, 17 Genes 423 (2026); Chiara Fantinato, Ane Elida Fonneløp, Øyvind Bleka, Magnus Dehli Vigeland & Peter Gill, The Invisible Witness: Air and Dust as DNA Evidence of Human Occupancy in Indoor Premises, 13 Sci. Rep. 19059 (2023); Chiara Fantinato, Peter Gill & Ane Elida Fonneløp, Investigative Use of Human Environmental DNA in Forensic Genetics, 70 Forensic Sci. Int’l: Genetics 103021 (2024); Marie Antony Dass et al., Assessing the Use of Environmental DNA (eDNA) as a Tool in the Detection of Human DNA in Water, 67 J. Forensic Sci. 2299 (2022); Int’l Org. for Standardization, The Mystery of the Phantom of Heilbronn (July 6, 2016); Claudia Himmelreich, Germany’s Phantom Serial Killer: A DNA Blunder, Time (Mar. 27, 2009); BBC News, “DNA Bungle” Haunts German Police (Mar. 28, 2009); Forensic Sci. Regulator (UK), Report on the DNA Contamination Case at LGC Forensics (Oct. 17, 2012; updated Jan. 3, 2025); Steven Morris, Rape Accused Was Victim of Forensics Error, Regulator Finds, Guardian (Oct. 1, 2012); Kristy A. Martire et al., Understanding “Error” in the Forensic Sciences: A Primer, 8 Forensic Sci. Int’l: Synergy 100470 (2024); Sci. Working Grp. on DNA Analysis Methods, Contamination Prevention and Detection Guidelines for Forensic DNA Laboratories (2017); President’s Council of Advisors on Sci. & Tech., Forensic Science in Criminal Courts: Ensuring Scientific Validity of Feature-Comparison Methods (Sept. 2016); John M. Butler et al., DNA Mixture Interpretation: A NIST Scientific Foundation Review, NISTIR 8351 (Dec. 2024); Nat’l Inst. of Just., When DNA Samples Are Complicated: Calculating Variation in Mixed Samples Interpretation (Dec. 12, 2022); William C. Thompson, Uncertainty in Probabilistic Genotyping of Low Template DNA: A Case Study Comparing STRMix™ and TrueAllele™, 68 J. Forensic Sci. 1049 (2023); Michael D. Edge & Jeanna Neefe Matthews, Open Practices in Our Science and Our Courtrooms, 38 Trends Genetics 113 (2022); President’s Council of Advisors on Sci. & Tech., An Addendum to the PCAST Report on Forensic Science in Criminal Courts (Jan. 2017); Nat’l Rsch. Council, Strengthening Forensic Science in the United States: A Path Forward (2009); Eric S. Lander, Fixing Rule 702: The PCAST Report and Steps To Ensure the Reliability of Forensic Science, 86 Fordham L. Rev. 1661 (2018); Adam B. Shniderman, Prosecutors Respond to Calls for Forensic Science Reform: More Sharks in Dirty Water, 126 Yale L.J. Forum 348 (2017); U.S. Dep’t of Just., National Commission on Forensic Science, Work Products; Nat’l Comm’n on Forensic Sci., Views of the Commission: Testimony Using the Term “Reasonable Scientific Certainty” (Mar. 22, 2016); U.S. Dep’t of Just., Attorney General Jeff Sessions Announces New Initiatives To Advance Forensic Science and Help Counter the Rise in Violent Crime (Apr. 10, 2017); Nat’l Comm’n on Forensic Sci., Meeting #9 Minutes (Mar. 21–22, 2016); John M. Butler et al., NIST Interlaboratory Studies Involving DNA Mixtures (MIX05 and MIX13): Variation Observed and Lessons Learned, 37 Forensic Sci. Int’l: Genetics 81 (2018); Steve Horn, U.S. Government Lab Withheld Groundbreaking Study for 5 Years That Can Help Defendants Question Reliability of Certain DNA Evidence, Criminal Legal News (Mar. 2019); Expert Working Grp. on Human Factors in Forensic DNA Interpretation, Forensic DNA Interpretation and Human Factors: Improving Practice Through a Systems Approach, NIST IR 8503 (May 2024); Donald E. Shelton, The “CSI Effect”: Does It Really Exist?, NIJ J., No. 259 (Mar. 2008); Logan A. Ewanation, Susan Yamamoto, Jordan Monnink & Evelyn M. Maeder, Perceived Realism and the CSI-Effect, 3 Cogent Soc. Scis. 1294446 (2017); Defense Motion in Limine #6 Re: Rylene Nowlin and “Touch” and “Contact” DNA, State v. Kohberger, No. CR01-24-31665 (4th Jud. Dist., Ada Cnty., Idaho Feb. 24, 2025); Order on Defendant’s Motion in Limine Re: Rylene Nowlin and “Touch” and “Contact” DNA, State v. Kohberger, No. CR01-24-31665 (4th Jud. Dist., Ada Cnty., Idaho Apr. 18, 2025); Mike Ware & Angie Ambers, The Wrongful Conviction of Lydell Grant for Murder in Texas: A Cautionary Tale of Faulty Eyewitness Identification and the Perils of Inconclusive DNA Mixture Statements, ISHI Report (May 2021); Innocence Project of Tex., Lydell Grant; Ex parte Grant, 622 S.W.3d 392 (Tex. Crim. App. 2021); Casey J. Bastian, Texas Man Positively Identified by Six Eyewitnesses and Sentenced to Life Granted Actual Innocence Relief as Result of DNA Evidence, Criminal Legal News (Dec. 2021); Nat’l Inst. of Just., The Impact of False or Misleading Forensic Evidence on Wrongful Convictions (Nov. 28, 2023); John S. Morgan, Wrongful Convictions and Claims of False or Misleading Forensic Evidence, 68 J. Forensic Sci. 908 (2023); Marcellus Williams Is Executed Despite Prosecutors and the Victim’s Family Asking That He Be Spared, CNN (Sept. 25, 2024); Missouri Executes Marcellus Williams Despite Questions Over Evidence, After Supreme Court Denies Final Bid for Delay, CBS News (Sept. 25, 2024); Innocence Project, Man Faces Execution on Sept. 24 Despite Evidence of Innocence; Death Penalty Info. Ctr., Three U.S. Supreme Court Justices Decry “Inexplicable” Texas Refusal To Test DNA in Rodney Reed Case (Mar. 25, 2026); Supreme Court Rejects Appeal from Texas Death Row Inmate Rodney Reed, Tex. Trib. (Mar. 23, 2026); Death Penalty Info. Ctr., The Limitations of DNA Evidence in Innocence Cases (Oct. 23, 2024; updated Mar. 14, 2025); Christi J. Guerrini et al., Four Misconceptions About Investigative Genetic Genealogy, 8 J.L. & Biosciences lsab001 (2021); Christi J. Guerrini et al., Investigative Genetic Genealogy Practices Warranting Policy Attention: Results of a Modified Policy Delphi, 21 PLOS Genetics e1011520 (2025); U.S. Dep’t of Just., Interim Policy: Forensic Genetic Genealogical DNA Analysis and Searching (2019); Brendan I. Koerner, Your Relative’s DNA Could Turn You Into a Suspect, Wired (Oct. 13, 2015). 

Subscribe today

Already a subscriber? Login