In the Iowa Supreme Court No. 24–1123
Submitted September 9, 2025—Filed May 29, 2026 Donald Lee Wyldes Jr.,
Appellant,
vs.
State of Iowa,
Appellee.
Appeal from the Iowa District Court for Wayne County, Dustria A. Relph (motion to quash) and Elisabeth Reynoldson (summary judgment and trial), judges.
A defendant appeals the denial of his postconviction-relief action challenging, among other things, firearm toolmark expert testimony introduced at his trial. Affirmed.
McDermott, J., delivered the opinion of the court, in which all justices joined.
Erica A. Nichols Cook (argued) and Elaina Steenson, of the Wrongful Conviction Unit, State Public Defender’s Office; M. Chris Fabricant and Tania Brief of The Innocence Project, New York, New York; and Megan Richardson of The Exoneration Project, Chicago, Illinois, for appellant.
Brenna Bird, Attorney General, and Louis S. Sloven (argued), Assistant Attorney General, for appellee.
Jesse Linebaugh and Joseph R. Quinn of Faegre Drinker Biddle & Reath LLP, Des Moines, for amicus curiae The Innocence Network.
Matthew Sease, Des Moines; Donald P. Salzman, Washington, D.C.;
Marley Ann Brumme, Boston, Massachusetts; and Hannah Henderson, Boston, Massachusetts, for amici curiae Criminal Law Scholars, Scientists and Statisticians.
3
McDermott, Justice.
Donnie Lee Wyldes Jr. was convicted of first-degree murder and attempted murder in 1987. He appeals the denial of his third application for postconviction relief, in which he argues that new scientific evidence invalidates the State’s forensic testimony used to convict him. In particular, Wyldes challenges the reliability of firearm toolmark and shoe print evidence introduced at his trial, citing modern studies that he claims label these methods fundamentally flawed. The district court denied his application. Wyldes asks us to reverse that decision and grant him a new trial.
I.
A. The Crime. On October 15, 1986, night settled on Ronald and Ruby Starnes’s farmhouse outside Corydon like most any other Wednesday. The couple had lived there since 1952. As Ruby read quietly in bed, sounds from the 10 p.m. news played in the next room; Ronald rarely missed a broadcast.
The quiet evening was interrupted by an unfamiliar clatter that caught Ronald’s ear. Puzzled, he walked to the bedroom to ask Ruby if she had heard it. She had; it sounded to her like something on the roof. Ronald, a sixty-six-year- old retired farmer, grabbed a flashlight, put on a shirt and cap, and stepped out into the night to investigate.
Soon after, Ruby heard a series of loud pops that sounded like firecrackers, followed by a heavy thud against the side of the house. Thinking Ronald was trying to scare a raccoon off the roof, she shouted out the window for him to stop hitting the house so hard. No one answered.
Uneasy, Ruby went to the kitchen and opened the porch door. A short man, about 5'2", dressed in dark clothing, was walking up the steps, his dark
4
brown eyes visible behind a brown mask pulled low over the rest of his face and neck. This wasn’t Ronald.
Ruby slammed the door, but before she could lock it, the man forced it partially open. As Ruby pushed back, bracing herself against the door with all her strength, more pops rang out—and this time, there was no mistaking the gunshots for firecrackers. Ruby screamed that the sheriff was already on the way, and the intruder finally fled.
The farmhouse fell into a heavy silence as Ruby waited to make sure he had really gone. After several minutes, Ruby tried to call 911, but the phone line was dead—the wires had been cut. When she left the house to try to flag down a car for help, she found Ronald. He was lying face down beneath a tree, shot eight times. He was dead.
B. The Investigation. The police investigation initially had little to go on.
Police found no murder weapon and no fingerprints—only a partial shoe print, bullets, and ten spent .22 caliber shell casings. The casings were scattered around: one on the porch, six near Ronald’s body, and three on the gravel road roughly two-tenths of a mile from the farmhouse. Investigators also noted that the phone lines had been cut and the Starnes’s tires had been slashed, indicating a deliberate and premeditated ambush.
The investigation soon turned toward Wyldes. Four days before the murder, Wyldes had slid his car into a ditch near the farmhouse during a heavy rainstorm. He had walked to the Starnes’s home to ask for help, and the couple had kindly let him inside to use their phone.
When interviewed three days after the murder, Wyldes acknowledged the prior visit but provided an alibi for the night of the murder. He claimed he was working on a trailer with two friends, Jay Kanney and Bobby Easley, until
5
around 10 p.m. Wyldes claimed he drove Kanney home at that point and then remained there for half an hour before heading to his grandparents’ house for the evening, where he arrived at around 11 p.m.
But both Kanney and Easley told police a different account of the events.
Both said that Easley—not Wyldes—had driven Kanney home that night. Easley said he stopped to buy beer after dropping Kanney home, and he arrived at his own home by 10:30 p.m. Kanney’s wife corroborated this, stating that she heard Easley’s truck arrive and that she never saw Wyldes that night, despite his claim that he spent half an hour inside her home.
In January 1987, as the investigation continued, Kanney called investigators to inform them about a conversation with Wyldes on the day Wyldes’s car slid into the ditch near the Starnes’s farmhouse. Kanney had picked Wyldes up from the Starnes home after Wyldes walked there to use the phone. On the drive, Wyldes told Kanney that after his car went into the ditch, he fired a gun from the road to make sure it worked in case he came across any mean dogs in the area. This information was significant because it potentially explained the shell casings found on the gravel road—a detail Kanney would not have known. Kanney also reported seeing Wyldes carrying a Marlin-brand .22 caliber rifle on a strap slung over his shoulder both the day he picked up Wyldes and the next when they returned to tow Wyldes’s car from the ditch. Kanney recalled Wyldes placing the rifle in his car afterward. Both Kanney and Easley believed they had seen Wyldes with the same rifle after the murder.
With this new information, police reinterviewed Wyldes. Wyldes’s story shifted repeatedly. Although Wyldes had previously told police that he didn’t own any firearms, he now admitted to owning firearms at various points. When asked to list those firearms, he omitted mention of the Marlin. He later admitted to
6
buying the Marlin but claimed he had pawned it and never bought it back. Over the course of several later interviews, he admitted to having the Marlin with him when he walked to the farmhouse to use the phone days before the murder, but he denied firing it on the road. He also claimed that after Kanney picked him up, he left the Marlin in Kanney’s vehicle, and that Kanney later told him it had been stolen from the vehicle.
The murder weapon became a focal point of the State’s investigation. The State sought to use “firearm toolmark” analysis—the practice of examining markings on bullets or casings—to determine the firearm used in the crime. Robert Harvey, a certified firearm toolmark examiner with the Iowa Division of Criminal Investigation, concluded that the recovered bullets had been fired from a Marlin .22 caliber rifle. In analyzing the shell casings, Harvey determined that all ten shell casings found at the farmhouse and on the gravel road were fired from the same rifle. He based this conclusion on a characteristic common to each casing: a deep gouge spanning the length of the cartridge case, which Harvey considered to be “very unique.”
Despite the rainstorm on the day Wyldes slid into the ditch, the casings found on the gravel road were neither dusty nor muddy. What’s more, the casings were found about 250 yards from where Wyldes went into the ditch, but in the opposite direction from the Starnes residence. Yet relying on Kanney’s account, investigators pursued the theory that the casings were left after Wyldes fired several shots after walking out of the ditch. To test the link between these casings and Wyldes, investigators collected samples from various locations where Wyldes had practiced shooting, including .22 caliber casings from his father-in-law’s farm in South Dakota, six casings from a makeshift shooting range near Seymour, and a staggering 18,000 casings from a range in Linn County. Harvey
7
determined that a subset of the casings bore similarities to the casings found at the crime scene but were not exact matches. Harvey proposed a theory of “progressive deterioration” of the rifle that fired them, where a mechanical flaw in the rifle was worsening with use, leaving a deeper gouge over time.
Based on all the evidence collected, four months after the crime, officers arrested Wyldes for the murder of Ronald and the attempted murder of Ruby.
C. The Trial. The State featured Harvey’s firearm toolmark testimony in its prosecution at trial. He testified with certainty that the seven casings found at the scene and the three found on the gravel road were fired from the same firearm. Harvey also presented his progressive deterioration theory to the jury. He noted statements from other witnesses about Wyldes’s tendency to “dry fire” the Marlin—pulling the trigger on the rifle when the chamber is empty. Repeated dry firing, according to Harvey, could hasten the progressive deterioration of a rifle’s firing pin and thus deepen markings on a cartridge.
Although Harvey acknowledged that he could not definitively say the casings gathered from the South Dakota, Seymour, and Linn County ranges were fired from the same gun, he testified that they could have been if the mechanical gouge grew worse over time, eventually resulting in the large markings found on the casings at the farmhouse and on the road. Harvey also explained the perceived rarity of these marks, informing the jury that because only 91 of 18,000 casings from the Linn County range had a similar gouge, the rate of occurrence was about 0.5%. As Harvey further explained, out of every 1,000 casings he examined, only 5 possessed a marking indicating a mechanical problem like the one on the rifle used at the scene of the crime.
Wyldes did not entirely dispute the validity of Harvey’s forensic analysis.
Instead, he presented testimony from his own firearm toolmark expert, John
8
Cayton. Regarding the casings found at the farmhouse and on the road, Cayton’s opinion aligned with Harvey’s that they were fired from the same firearm. But Cayton disagreed with Harvey’s theory of progressive deterioration. He also rejected Harvey’s conclusion that the casings recovered from other locations could have been fired from the same rifle as those found at the farmhouse or on the road.
The State also presented testimony from criminalist Frank Tarasi, a footprint analyst who compared partial prints found on the Starnes’s door to shoes seized from Wyldes. Tarasi testified that the prints were consistent with the defendant’s footwear, but stopped short of declaring a definitive match, noting that other shoes had the same size and tread design. To further undermine the defense, the State highlighted discrepancies in Wyldes’s account of the shoes; although he initially claimed he received them as a gift two months after the murder, he later admitted to owning them a full year before the crime occurred. Wyldes asked his own expert, Cayton, to opine about the shoeprints, and he generally agreed with Tarasi’s assessment.
Wyldes took the stand in his own defense. He attempted to explain the inconsistencies in his statements to officers by claiming he had misunderstood the investigators’ questions. He stated that he failed to mention his Marlin rifle because he believed officers were only asking about weapons currently in his possession at the time of the interview. Regarding the timeline of his shoe ownership, he testified that he interpreted the phrase “last Christmas” as the holiday preceding the murder rather than the one that had just passed. Wyldes attributed his initial evasiveness to nervousness, fearing the legal ramifications if a rifle previously stolen from him had been used in a crime. And he argued that his overall cooperation, including directing investigators to locations where
9
they could find shell casings from his rifle, showed that his inconsistent statements were merely innocent misunderstandings.
During closing arguments, the prosecution relied heavily on Harvey’s forensic analysis to link Wyldes’s Marlin rifle directly to the crime scene. The State argued that there was “no question that the casings located at the Starnes’ residence and also up on the gravel road were identical” and were “definitely shot from the same gun.” The State emphasized the “very unique” nature of the casings, describing them as being “completely split out.” The State pointed out that even Cayton, the defense’s own expert, had corroborated these findings by testifying that the casings from both the residence and the gravel road were “identical, fired from the same gun, no question about it.”
The jury returned guilty verdicts on both charges. The district court sentenced Wyldes to life in prison without the possibility of parole for the first- degree murder of Ronald and to an additional twenty-five-year sentence for the attempted murder of Ruby.
D. Direct Appeal and Other Challenges. Wyldes appealed. The court of appeals affirmed both convictions, and we denied Wyldes’s application for further review. Procedendo was issued in 1989. Wyldes filed an application for postconviction relief in 1991 and a second application in 2007. He also sought relief through applications for writs of habeas corpus in 1994 and 2010. Each application was denied, and each appeal was unsuccessful.
Wyldes filed this postconviction-relief application, his third, in late 2010.
It sat dormant for a decade until he moved to amend the filing in 2020. The district court granted the motion, allowing Wyldes to assert six distinct grounds for relief. His claims included newly discovered evidence regarding forensic ballistics and shoeprint examinations, an alleged Brady violation, ineffective
10
assistance of counsel during both his trial and second postconviction-relief cases, improper jury instructions, and jury misconduct. The State filed a motion for summary judgment, arguing that all the claims were time-barred.
The district court granted the State’s motion in part, finding that all but the newly discovered evidence claim were untimely. After a hearing on the new evidence claim, the district court rejected Wyldes’s challenge and denied his request for a new trial. Wyldes appeals that order.
II.
A person convicted of a crime can challenge their conviction or sentence after their direct appeal has ended by filing an application for postconviction relief. Davis v. State, ___ N.W.3d ___, ___, 2026 WL 1261001, at *1 (Iowa May 8, 2026). Direct appeals typically focus on judicial errors made during trial, but an application for postconviction relief is a civil proceeding that permits the defendant to raise certain issues that weren’t or couldn’t have been taken up in the direct appeal. Id. at ___, 2026 WL 1261001, at *1; see also Iowa Code § 822.2(1)(d) (2011). Postconviction-relief proceedings are governed by Iowa Code chapter 822.
Iowa Code § 822.2(1)(d) allows a convicted person to seek relief if “[t]here exists evidence of material facts, not previously presented and heard, that requires vacation of the conviction or sentence in the interest of justice.” To succeed under this provision, an applicant must show that the new evidence (1) was discovered after the verdict, (2) could not have been discovered earlier through proper due diligence, (3) is material rather than cumulative or impeaching, and (4) “probably would have changed the result of the trial.” Jones v. State, 479 N.W.2d 265, 274 (Iowa 1991). We have described the probably-would-have-changed-the-result standard as “a high one” in light of “the
11
interest in bringing finality to criminal litigation.” More v. State, 880 N.W.2d 487, 499 (Iowa 2016).
A. Firearm Toolmark Evidence. Wyldes argues that although firearm toolmark analysis was considered sound at the time of his trial, scientific and statistical advancements have revealed it to be unreliable and thus unfit for expert testimony. He raises a similar challenge to footwear impression analysis, arguing that the methods presented during his 1986 trial are now outdated and have led to conclusions that the scientific community has since rejected. The State disputes these claims, asserting that Wyldes’s criticisms lack support and that recent data reinforces the validity of both forensic techniques.
1. The postconviction-relief hearing. During the postconviction-relief hearing, Wyldes introduced a variety of studies, reports, and expert testimony intended to challenge the scientific foundations of firearm toolmark and shoeprint examinations. The centerpiece of this evidence focused on firearm toolmark analysis and featured testimony from William Tobin, a forensic metallurgist specializing in the validity and reliability of forensic practices. Tobin detailed several critiques of the methodology used by the Association of Firearm Toolmark Examiners (AFTE), maintaining that these methods are unsupported by contemporary scientific literature.
Tobin testified that the process of comparing marks on cartridges or bullets lacks scientific credibility for several reasons. First, he argued that the comparison method relies on a presumption of uniqueness that has never been proven. Second, he points out that while the AFTE allows for a “same source” identification when marks show “sufficient agreement,” this standard lacks a precise definition and thus effectively leaves the final determination to an examiner’s subjective judgment rather than objective data. Finally, Tobin
12
identified a critical flaw—which he referred to as the “900-pound gorilla in the room”—concerning the inability of firearm toolmark examiners to reliably distinguish between characteristics shared by an entire manufacturing line of firearms and characteristics truly unique to an individual firearm.
Tobin’s final point addressed three distinct categories that forensic examiners use to classify toolmarks. At the broadest level are class characteristics, referring to marks common to a large group of firearms based on a manufacturer’s design specifications. Below these are subclass characteristics, referring to marks created by the same equipment at roughly the same time during a specific manufacturing batch. At the most granular level are individual characteristics, which are marks that are unique to a single firearm rather than a group.
Tobin argued that the same-source identification offered by Harvey and Cayton during the 1987 trial was improper. He contended that the most “scientifically and forensically defensible” conclusion those experts could have reached would have been to state that Wyldes’s rifle could have been or was possibly the source of the cartridge cases. According to Tobin, although an examiner can accurately identify class-level marks because different gun models have known, distinct features, the same reliability doesn’t exist at the subclass or individual levels. Without examining every firearm in a specific production line, an expert cannot truly know whether a mark is unique to one gun or simply a shared characteristic of that entire manufacturing batch.
The State countered Tobin’s testimony in large part by relying on a 2020 report from the Ames National Laboratory (in Ames, Iowa) known as the “Ames- II” study. See Stanley J. Bajic et al., Report: Validation Study of the Accuracy, Repeatability, and Reproducibility of Firearm Comparisons (2020). Ames-II was
13
developed in response to a 2016 report from the President’s Council of Advisors on Science and Technology (PCAST), which criticized earlier forensic toolmark research as structurally flawed and insufficient to establish scientific validity. The study was designed to satisfy PCAST’s rigorous standards, measuring how accurately and consistently certified examiners could link casings or bullets to a specific firearm. The study was intentionally designed to present a demanding test of examiner proficiency.
The results of the Ames-II study separately categorized the analysis of bullets and cartridges, although both evidence types ultimately yielded comparable results. During the testing process, examiners were required to sort pairs of evidence into three distinct categories based on their observations: “identification” was used when the examiner determined the pair was definitively fired by the same firearm, “elimination” was used when it was clear the items came from different weapons, and “inconclusive” was used in cases where available markings provided insufficient data to make a firm determination. An inconclusive result had three subcategories for an examiner to choose: (1) some agreement of individual characteristics but not enough for an identification, (2) neither agreement nor disagreement of individual characteristics, and (3) disagreement of individual characteristics, but not enough for an elimination.
Wyldes and the State presented two fundamentally different interpretations of what an “inconclusive” finding signified. Tobin argued that since each pairing had a correct answer (certainly the test designers knew whether a pair matched), any inconclusive response should be categorized as an error and counted against the examiner’s accuracy score. The State countered that it is a good thing for an examiner to admit when evidence is insufficient rather than guessing at an answer. Under the State’s logic, the inconclusive
14
responses represent a valuable display of professional restraint and should be excluded from error rate calculations.
These two interpretive approaches result in vastly different results about reliability. Under the State’s view, which excludes inconclusive answers, the Ames-II study shows a cartridge error rate of only 0.92% elimination to 1.76%. Under Wyldes’s view, the same data set causes the reported error rate to grow to between 25.6% and 51.5%. A similar discrepancy appears in the analysis of bullet evidence, demonstrating that the perceived validity of the science depends almost entirely on how these “inconclusive” findings are categorized. The results are summarized in these tables:
Ames II Study Correct answer was Correct answer was (cartridges, round 1) identification elimination (n=1,420) (n=2,835)
Correct answer 1,056 (74.37%) 1,375 (48.50%)
Inconclusive 339 (23.87%) 1,434 (50.58%)
Incorrect answer 25 (1.76%) 26 (0.92%)
Ames II Study Correct answer was Correct answer was (bullets, round 1) identification elimination (n=1,405) (n=2,842)
Correct answer 1,076 (76.60%) 961 (33.80%)
Inconclusive 288 (20.48%) 1,861 (65.50%)
Incorrect answer 41 (2.92%) 20 (0.70%)
The Ames-II study also evaluated repeatability, which measures an individual examiner’s ability to reach the same conclusion when presented with the same evidence later. For bullets, examiners agreed with their own previous conclusions 79% of the time for same-source (i.e., fired by the same firearm) samples, and 64.7% for different-source (i.e., fired from different firearms) samples. Cartridge analysis showed comparable results, with examiners agreeing with their own prior findings 75.6% of the time for same-source samples and 62.2% for different-source samples. The study’s authors interpreted these
15
figures as evidence of high repeatability, noting that the results exceeded the statistically expected levels of agreement “by a fairly wide margin.” They further clarified that most internal discrepancies involved subtle shifts between “inconclusive” subcategories rather than hard changes, such as switching a definitive identification to a definitive elimination.
Tobin, conversely, characterized these findings as “disturbing.” He was particularly bothered by the Ames-II study’s data on “reproducibility,” which measures how often two different examiners arrive at the same conclusion when viewing the same evidence. In the Ames-II study, two examiners both correctly identified a same-source bullet pair 67.8% of the time, and both correctly eliminated different-source bullets 30.9% of the time. For cartridges, the rates of mutual agreement were 63.6% for identifications and 40.3% for eliminations. The Ames-II authors maintained that these numbers were satisfactory because they fell above the baseline for expected agreement. Tobin asserted the opposite, citing these figures as proof that the practice is inherently subjective and lacking a consistent, reliable process that produces uniform results.
2. Cases addressing firearm toolmark examinations. In the years since the PCAST report’s critique of forensic toolmark examinations, courts across the country have scrutinized the validity of this evidence. Many of these cases address whether firearm toolmark evidence should be admitted during the initial trial. See, e.g., United States v. Pete, No. 3:22cr48, 2023 WL 4928523 (N.D. Fla. July 21, 2023); United States v. Romero-Lobato, 379 F. Supp. 3d 1111 (D. Nev. 2019); Abruquah v. State, 296 A.3d 961 (Md. 2023). The legal posture in this case is different, with Wyldes’s challenge raised in a postconviction-relief action based on the discovery of new evidence after a conviction several decades ago.
16
In United States v. Romero-Lobato, a Nevada federal district court evaluated the reliability of firearm toolmark expert testimony by applying the five Daubert factors. 379 F. Supp. 3d at 1118–22. These criteria examine whether a scientific method (1) is testable, (2) has been subjected to peer review and publication, (3) has a known or potential rate of error, (4) includes standards controlling its operation, and (5) enjoys general acceptance within the relevant community. Id. The Romero-Lobato court found that the testability, peer review and publication, error rate, and general acceptance factors all weighed in favor of admitting the firearm toolmark testimony, but concluded that the fourth element, existence of controlling standards, weighed against admissibility. Id. at 1122. Ultimately, the judge determined that the balance tipped strongly in favor of admitting the testimony. Id. The court further emphasized that it found no federal precedent for a total ban on firearm identification testimony and was unwilling to take such a drastic step without “a remarkable argument supported by remarkable evidence.” Id.
In United States v. Pete, a Florida federal district court conducted a similar Daubert analysis and concluded that the only factor of concern was the lack of objective controlling standards. 2023 WL 4928523, at *4–6. Much like the court in Romero-Lobato, it determined that the concerns surrounding subjectivity were outweighed by low error rates and other factors. Id. Notably, the Pete court considered the examiner’s personal experience as a key component of reliability. Id. at *5. It reasoned that although inexperienced examiners might produce an outsized number of errors, the testimony of a highly experienced professional should be granted a greater assumption of accuracy. Id. Based on this logic, the court permitted the firearm toolmark testimony without imposing limitations. Id.
17
In United States v. Williams, a federal court in Illinois rejected a Daubert challenge, concluding that a growing body of research continues to reinforce the field’s underlying principles. No. 19–CR–00282, 2025 WL 2444098, at *7–8 (N.D. Ill. Aug. 25, 2025). The court emphasized that it had seen no evidence suggesting firearm toolmark analysis suffers from anything other than a consistently low error rate. Id. at *8. Similarly, in United States v. Lang, a federal court in Florida, relying on Pete, declined to force a firearm toolmark expert to qualify their findings. No. 2:19–cr–150–SPC–NPM, 2025 WL 1921085, at *1–2 (M.D. Fla. July 11, 2025). The court was satisfied that any concerns regarding the expert’s degree of certainty could be adequately addressed through cross-examination and competing expert testimony. Id. at *2.
A divided Maryland Supreme Court took a more restrictive approach in Abruquah v. State, ruling that firearm toolmark expert testimony may no longer include unqualified conclusions that definitively match a bullet or cartridge to a specific firearm. 296 A.3d at 998. Using an expanded version of the Daubert analysis, the majority identified what it deemed to be several failures in the discipline, including unreliable error rates, a lack of objective standards, and an “analytical gap” between the underlying science and the definitive claims made by experts. Id. at 988–97.
The Abruquah majority opinion contained extensive analysis of how inconclusive responses are used to mask potential errors. Id. at 989–91. The court challenged the prosecution’s argument—similar to the one made here— that an inconclusive result is a scientifically “correct” answer reflecting professional caution. Id. The court pointed instead to a discrepancy between study designs: in “closed-set” studies, where a match is guaranteed to exist, examiners almost never choose the inconclusive option, but in “open-set” studies
18
such as Ames-II, where a match is not guaranteed, inconclusive responses occur at a much higher frequency. Id. at 989–90. The majority also noted that accuracy appears inconsistent across different brands of firearms; although examiners correctly matched 89.7% of Beretta handguns, for instance, their accuracy fell to 56.6% for Ruger handguns. Id. at 990.
One of two dissenting opinions in Abruquah contended that inconclusive determinations are irrelevant to the court’s purposes, and that the accuracy of determinations only matters “when the examiner does declare an identification or elimination.” Id. at 1007 (Gould, J., dissenting). In other words, if an examiner is consistently correct whenever they state an identification or elimination, the methodology remains a valid tool for trial. Id. The dissenting justice was satisfied with the current data on repeatability and reproducibility, arguing that false- positive rates remain well below the thresholds recommended even by PCAST and that firearm toolmark experts should thus be permitted to offer unqualified testimony. Id. at 1016.
3. The “new evidence” claim. Wyldes argues that the recent studies undermining the reliability of firearm toolmark analysis constitute new evidence entitling him to a new trial under Iowa Code § 822.2(1)(d). In More v. State, we declared that significant scientific developments, including contemporary studies providing “new statistical data previously unavailable to scientists,” could qualify as new evidence for postconviction-relief claims if other criteria were met. 880 N.W.2d at 509–10. Again, to prevail on a new evidence claim, a postconviction relief applicant must establish four elements:
(1) that the evidence was discovered after the verdict; (2) that it could not have been discovered earlier in the exercise of due diligence; (3) that the evidence is material to the issues in the case and not merely cumulative or impeaching; and (4) that the evidence probably would have changed the result of the trial.
19
Jones, 479 N.W.2d at 274. Applying More, we agree that Wyldes satisfies the first two elements, as the firearm toolmark studies since his 1987 trial constitute evidence “discovered after the verdict” and “could not have been discovered earlier.” 880 N.W.2d at 499 (quoting Jones, 479 N.W.2d at 274).
In More, we also found that new evidence was material and not merely cumulative or impeaching, even when the defendant’s trial expert had criticized the forensic process the defendant would later attack again in the postconviction-relief action. Id. at 509–10. We distinguished between simply “advancing the state of the art” and a “blockbuster report” that enabled a significantly deeper critique than what was available at trial. Id. at 509. Wyldes’s expert, Cayton, not only didn’t attack the validity of firearm toolmark analysis at trial, he embraced it. Although he disputed some of the State’s specific findings, Cayton’s criticisms focused on the results rather than the methodology itself. Cayton corroborated one of the State’s most damaging claims by agreeing that the casings found on the gravel road and at the farmhouse originated from the same rifle. As a result, Wyldes also establishes the third element, as the new evidence that he now provides is not cumulative or impeaching but a new line of attack that was never presented to the jury.
The final requirement asks whether the new evidence probably would have changed the trial’s result. As established in More, this is not a “harmless error” assessment, but a case-specific, fact-intensive inquiry into whether the new evidence would have prompted the jury to reach a different verdict. Id. at 510. Because this analysis is backward-looking, we evaluate how the original 1987 trial would have been altered if this information had been available. This involves determining the specific effect the new data would have had on the expert
20
testimony presented at that time, including whether that testimony would have remained admissible.
The methodology performed by both firearm toolmark experts at the original trial is fundamentally the same as the techniques tested in the Ames-II study and other recent research cited by courts. These studies demonstrate that the practice maintains relatively low error rates that fall well within statistically acceptable margins. What’s more, the research confirms that the metrics for reproducibility and repeatability similarly suggest the method’s reliability. While the scientific community is obviously not unanimous in its endorsement, the results from numerous large-scale, rigorous studies suggest that forensic toolmark examination continues to enjoy general acceptance.
From our review of the cases, it appears that most courts have concluded that firearm toolmark testimony generally remains admissible in court and enjoys widespread acceptance in the field among examiners. See, e.g., United States v. Brown, 973 F.3d 667, 704 (7th Cir. 2020) (affirming a district court that described firearm toolmark analysis as “widely accepted beyond the judicial system”); United States v. Richardson, No. 19–20076, 2024 WL 961228, at *5–9 (D. Kan. Mar. 6, 2024) (noting that “the field, as well as the academy, has and continues to appropriately respond with more studies and more stringently designed studies of this important area of forensic science,” which “still enjoys general acceptance among firearms examiners”); United States v. Graham, No. 4:23–cr–00006, 2024 WL 688256, at *13 (W.D. Va. Feb. 20, 2024) (“Toolmark identification enjoys widespread acceptance around the world.”). While the testimony provided at the postconviction-relief hearing and the decisions in cases such as Abruquah offer insightful critiques, we agree with the majority of
21
courts in rejecting Wyldes’s argument on the inadmissibility of firearm toolmark testimony.
We are unpersuaded by the Abruquah majority’s rejection of firearm toolmark evidence in part based on the differing standards that Maryland and Iowa courts apply in considering the admissibility of expert testimony. The court in Abruquah applied Maryland’s more stringent standard (referred to as the “Daubert-Rochkind” test) for evaluating scientific evidence under Maryland’s rules of evidence. 296 A.3d at 971–72. That test requires courts to consider a nonexhaustive list that includes no fewer than ten different factors to determine whether the expert testimony is sufficiently reliable. Id. Iowa, on the other hand, has generally taken a more liberal view on the admissibility of expert testimony. See Hutchison v. Am. Fam. Mut. Ins., 514 N.W.2d 882, 885 (Iowa 1994). We apply a simpler two-part inquiry to determine the admissibility of expert testimony. Ranes v. Adams Lab’ys, Inc., 778 N.W.2d 677, 685 (Iowa 2010). First, the court must “determine if the testimony ‘will assist the trier of fact’ in understanding ‘the evidence or to determine a fact in issue.’ ” Id. (quoting Iowa R. Evid. 5.702). Second, the court must evaluate whether “the witness is qualified to testify ‘as an expert by knowledge, skill, experience, training, or education.’ ” Id. (quoting Iowa R. Evid. 5.702). Although we have approved of courts considering the Daubert factors in certain types of cases, see Leaf v. Goodyear Tire & Rubber Co., 590 N.W.2d 525, 532 (Iowa 1999) (en banc), we have never adopted or required the stricter Daubert test, Ranes, 778 N.W.2d at 685–86.
As the district court noted, Wyldes thus has the challenging task of overcoming a low bar for admissibility of expert testimony, on the one hand, and the heavy burden of proving that the outcome at trial probably would have been different, on the other. Wyldes’s best argument is that a forensic examiner who
22
doesn’t possess the source firearm will almost never be able to state with absolute certainty, as Harvey did, that two samples originated from the same firearm to the exclusion of all others. Assuming Wyldes made full use of this new evidence and essentially presented Tobin’s testimony from the postconviction- relief hearing, his expert would no longer agree with the State that the casings on the gravel road and at the farmhouse bore similar markings that could have come from the same rifle.
But the central question remains whether this new information “probably would have changed the result of the trial.” Jones, 479 N.W.2d at 274. We have already concluded that firearm toolmark evidence remains admissible in Iowa, and we do not believe that a jury armed with the results of the Ames-II study would, in fact, find the method wholly unreliable. Even if we excised the testimony about the “certainty” of a match, a jury hearing the bulk of Harvey’s testimony and the other evidence in the case would, in our view, reach the same verdict. Harvey could still testify that the casings found at the scene originated from a Marlin .22 caliber rifle, a relatively simple class-level identification based on design specifications. He could also explain that the casings from the gravel road and the farmhouse shared visual similarities not produced by every Marlin, making it impossible to rule out the rifle as the same source. Harvey could also present the progressive deterioration theory; although Wyldes would attack the theory, that attack already played out in the original trial, and the jury sided with Harvey.
But that’s not all. In More, unlike here, we concluded that sufficient evidence supported the defendant’s conviction even though the challenged forensic methodology was deemed entirely useless. 880 N.W.2d at 510–11. Although the remaining evidence was purely circumstantial, it was convincing.
23
See id. The defendant had a clear motive: to cash in on the life insurance policy of the victim, his girlfriend. Id. at 510. He received a traffic ticket while speeding away shortly after the murder, placing him near the crime at the key time. Id. at 511. He was found with a bullet in his pocket of the same type used in the crime, for which he had no explanation. Id. He provided suspicious answers to investigators. Id. After his girlfriend’s murder, he left the state with his ex-wife, skipped the victim’s funeral, stayed on the run despite knowledge of a warrant out for his arrest, and was finally arrested in Montana. Id. at 495, 511. “[T]he combination of facts and circumstances,” we concluded, “strongly point toward [the defendant]’s guilt.” Id. at 511.
Applying that same logic here, even if the firearm toolmark testimony were circumscribed, the other evidence against Wyldes remains substantial. Wyldes initially denied owning any Marlin rifles. He later gave an unconvincing explanation for that denial when he admitted to owning one. Kanney saw Wyldes with the rifle shortly before the date of the murder. Although Wyldes claimed he left the rifle in Kanney’s vehicle and it was thereafter stolen, witnesses testified to seeing Wyldes with the rifle after the murder. His alibi for the night of the crime was directly contradicted by every person he claimed to have been with. He also provided misleading statements about when he acquired a pair of shoes that investigators were curious about, claiming that he’d received them after the murder when in fact he’d received them before it. And even without a definitive ballistics match, the jury would still hear that Wyldes fired a rifle on a nearby road where investigators recovered casings, and that those casings matched the caliber and type found at the crime scene. Wyldes was also familiar with the farmhouse layout from his visit there mere days before, entering through the same porch door that the killer later tried to kick down. And finally, he perfectly
24
matched the physical description provided by the surviving victim—a strikingly short man, barely five feet tall, with dark brown eyes.
Stated simply, we cannot conclude that a different verdict would probably have been rendered even with the introduction of Wyldes’s new evidence. Viewed in its entirety, the nonforensic evidence strongly indicates Wyldes’s guilt, and this conclusion is obviously reinforced by the firearm toolmark evidence that would remain admissible even if we were to excise the challenged testimony of a “certain” match. The evidence pointing toward Wyldes’s guilt is simply too significant to suggest that the jury’s decision probably would have changed had the recent studies been available in 1987. Because the trial’s outcome remains the most probable result regardless of the new evidence, Wyldes is not entitled to a new trial on these grounds.
B. Shoeprint Testimony. Wyldes also contends that modern developments in shoeprint analysis discredit the trial testimony linking prints found on the Starnes’s door to his own footwear. Wyldes argues that the original expert failed to account for a four-month lag between the murder and the seizure of the shoes. During the postconviction-relief hearing, testimony revealed that footwear examiners in the 1980s lacked standardized, discipline-specific methodologies and often relied on fingerprint analysis techniques instead. But unlike fingerprints, the outsole of a shoe—the part that leaves a print—is subject to significant wear and tear over time. A longer delay between the crime and the seizure of the evidence increases the likelihood that the shoe’s tread will have changed, making a comparison less reliable. Wyldes thus argues that the four- month gap renders the trial testimony, which identified his shoes as a potential match, entirely speculative.
25
Footwear comparison is a longstanding forensic practice widely accepted by courts. “The science of footwear analysis is neither new nor novel; expert testimony on footwear comparisons has been admitted in courts in the United States for years.” United State v. Mahone, 328 F. Supp. 2d 77, 89 (D. Me. 2004); see also United States v. Ross, 263 F.3d 844, 846–47 (8th Cir. 2001). Both the State’s and Wyldes’s experts acknowledged that a four-month time gap would be a critical factor in a modern forensic evaluation. Wyldes suggests that if a jury were presented with this context today, they would recognize that the outsoles could have changed and, as a result, would disregard the link to the crime scene.
But the original testimony provided by the State’s expert was limited. The State’s expert summarized his opinion this way: “[Wyldes’s] shoes could have made the prints that are present on [the farmhouse] door. However, I must also recognize that other shoes of the same type, the same size, and the same tread design do exist.” This testimony wasn’t an unqualified individual-level conclusion that Wyldes’s shoes made the prints, nor was it a subclass determination that the prints must have come from a particular manufacturing line of shoes that included Wyldes’s pair. This was a class-level conclusion that Wyldes’s shoes could have left the recovered shoeprint.
We find nothing in the postconviction-relief record to suggest that the original shoeprint testimony was inadmissible or fundamentally undermined by modern forensic standards. Notably, Wyldes’s own postconviction-relief expert arrived at essentially the same conclusion as the original trial expert, finding that even in light of the four-month lag, there was “an association of class characteristics.” In other words, Wyldes’s shoeprint expert agreed that Wyldes’s shoe could have left the print on the farmhouse door. Even if the defense had successfully used this new information to neutralize the original expert’s
26
findings, it is highly unlikely the trial’s outcome would have been different. Jurors would have heard that the testimony was perhaps less reliable due to the passage of time, but they would not have heard that Wyldes’s shoes could not have left the shoeprint. Neutralizing a mild conclusion of class similarity is not enough to overcome the weight of the other evidence in this case.
C. Due Process. Wyldes also asserts a due process claim. Due process under both the United States and Iowa Constitutions “requires fundamental fairness in a judicial proceeding.” More, 880 N.W.2d at 499 (quoting State v. Becker, 818 N.W.2d 135, 148 (Iowa 2012), overruled on other grounds by, Alcala v. Marriott Int’l, Inc., 880 N.W.2d 699 (Iowa 2016)). Evidence that is so unreliable that its admission creates fundamental unfairness violates due process. Id. at 500; see also Manson v. Brathwaite, 432 U.S. 98, 117 (1977). To trigger a due process violation, evidence must be more than “merely untrustworthy”—it must be “so inherently unreliable that to even allow a jury to consider it is a denial of due process.” State v. Bruns, 304 N.W.2d 217, 219 (Iowa 1981). This is a rigorous threshold designed to protect the integrity of the court, not to serve as a catchall for every piece of disputed testimony.
Wyldes argues that the presentation of “faulty forensic evidence” to the jury as conclusive proof of guilt violated his due process rights. His argument essentially repackages the firearm toolmark and shoeprint critiques under a constitutional label, arguing that the scientific testimony against him in 1987 has been thoroughly discredited by modern standards. He points us to authority establishing that false, debunked, and discredited scientific evidence can form the basis for a due process violation. See United States v. Ausby, 916 F.3d 1089, 1092, 1094–95 (D.C. Cir. 2019) (per curiam) (holding that the introduction of material “false” scientific evidence violates due process); State v. Bridges, No. 90
27
CRS 23102–04, 2015 WL 12670468, at *2 (N.C. Super. Ct. Oct. 1, 2015) (holding that debunked forensic evidence violated due process when it “overstated the significance of the hair analysis to the jury”); Ex parte Henderson, 384 S.W.3d 833, 834 (Tex. Crim. App. 2012) (per curiam) (remanding for a new trial where a conviction was obtained through subsequently discredited scientific evidence).
But having ruled as we have above about the admissibility of the State’s firearm toolmark and shoeprint evidence, we cannot characterize this evidence as totally false, debunked, or discredited. As mentioned, the rigorous Ames-II study demonstrated low error rates for firearm toolmark analysis within statistically acceptable margins, with reproducibility and repeatability results that similarly suggest the method’s reliability. Wyldes’s due process arguments about the shoeprint analysis similarly fail for the reasons explained above. “Fundamental unfairness is a high standard,” More, 880 N.W.2d at 512, and Wyldes cannot clear it. We thus affirm the denial of this claim.
D. Actual Innocence. Wyldes raised a claim of actual innocence below that the district court initially failed to rule on. Wyldes filed a motion to reconsider, but that motion was denied. Wyldes has thus preserved error on this claim under our standards, see Lamasters v. State, 821 N.W.2d 856, 858–59 (Iowa 2012), and we review his actual innocence claim de novo, Dewberry v. State, 941 N.W.2d 1, 4 (Iowa 2019).
To prevail on a claim of actual innocence, the applicant must show “that, despite the evidence of guilt supporting the conviction, no reasonable fact finder could convict the applicant of the crimes for which the sentencing court found the applicant guilty in light of all the evidence, including the newly discovered evidence.” Schmidt v. State, 909 N.W.2d 778, 797 (Iowa 2018). The applicant bears this burden under a “clear and convincing evidence” standard. Id. Wyldes
28
argues that the forensic evidence presented against him at trial was the State’s only direct evidence, and that without it, no reasonable factfinder could convict him.
We have already determined that Wyldes’s critiques of firearm toolmark and shoeprint analysis probably would not have changed the outcome at trial. This standard is lower than the clear and convincing evidence standard that Wyldes must overcome for his actual innocence claims. Given the strength of the circumstantial evidence and the firearm toolmark evidence remaining, even if we were to excise the challenged toolmark testimony of a definite match, Wyldes has not proven actual innocence. We thus affirm the denial of his actual innocence claim.
E. Summary Judgment Dismissing Claims for Untimeliness. Wyldes also appeals the district court’s granting of summary judgment of several other claims in his postconviction-relief application that were rejected as untimely filed.
Under Iowa Code § 822.3, postconviction-relief applications generally must be filed within three years of a conviction becoming final or the issuance of procedendo. The three-year time limit does not apply to claims that “could not have been raised within the applicable time period.” Id. To qualify for this exception, the applicant must demonstrate that the claim could not have been discovered through reasonable diligence. Jones, 479 N.W.2d at 274. The district court granted summary judgment on several of Wyldes’s claims, concluding that they were filed outside the three-year limit and failed to meet the criteria for the exception. We review each disputed ruling in turn.
1. Brady violation. Under Brady v. Maryland, the prosecution violates due process if it fails to disclose evidence favorable to the accused that is material to
29
either guilt or punishment. 373 U.S. 83, 87 (1963). Failing to disclose Brady material may result in a new trial. See State v. Barrett, 952 N.W.2d 308, 313 (Iowa 2020). Wyldes claims that the State committed a Brady violation by failing to disclose information about Harvey before the original trial. Harvey testified at trial that he had twenty years of firearm toolmark experience, while public records allegedly indicate he had only eight. Wyldes contends that this discrepancy was not discovered until 2022.
In response, the State argues that since the records in question were publicly available, they could have been discovered long before 2022. Under Iowa Rule of Civil Procedure 1.981(5), a party resisting summary judgment must “set forth specific facts showing that there is a genuine issue for trial.” See Slaughter v. Des Moines Univ. Coll. of Osteopathic Med., 925 N.W.2d 793, 808 (Iowa 2019) (requiring a party to produce actual evidence in resisting summary judgment rather than mere allegations). But Wyldes offered only a bare assertion that the evidence was unknown and undiscoverable until 2022. He offers no explanation as to why this information could not have been discovered decades earlier. Because Wyldes failed to show that he exercised reasonable diligence in discovering these public records, we affirm the district court’s grant of summary judgment on this issue.
2. Ineffective-assistance-of-counsel claims. Wyldes contends that his claim of ineffective assistance by his prior postconviction counsel should be exempt from the standard three-year limit under the exception established in Allison v. State, 914 N.W.2d 866, 888 (Iowa 2018). In Allison, we held that when a successive postconviction-relief application alleges that postconviction counsel was ineffective in presenting a claim of ineffective trial counsel, the second application can relate back to the filing date of the original if it’s filed promptly
30
after the first postconviction action concludes. Id. at 891. We later clarified that the relation-back rule only applies when the second application raises the exact same claim that the first attorney failed to handle effectively. Goode v. State, 920 N.W.2d 520, 526–27 (Iowa 2018). In 2019, the legislature amended § 822.3 to supersede Allison, providing that ineffective-assistance claims “shall not toll or extend the limitation periods . . . nor shall such claim relate back to a prior filing.” 2019 Iowa Acts ch. 140, § 34 (codified at Iowa Code § 822.3 (2020)); see also Sandoval v. State, 975 N.W.2d 434, 436 (Iowa, 2022) (noting the superseding of Allison).
Wyldes seeks to raise his Brady claim based on prior postconviction counsel’s ineffectiveness in failing to discover and raise it. Even if the 2019 amendments to § 822.3 do not apply, his attempt still stalls in two ways. First, the exception in Allison specifically addressed only a second postconviction-relief application, not a third application like the one Wyldes has filed here. See 914 N.W.2d at 891. Second, even if the exception were to apply to a third petition, the requirement for promptness is measured against the conclusion of the first postconviction proceeding. See id. Wyldes filed this application fourteen years after his first postconviction proceeding ended, failing the promptness test entirely. As a result, this ineffective-assistance claim is untimely, and we thus affirm the district court’s grant of summary judgment on this issue.
3. Subpoenas for other investigative files. After the district court dismissed several of his claims as time-barred, Wyldes served subpoenas for investigative files regarding what he characterized as similar crimes in areas neighboring Wayne County (where the crime occurred) around the time of the crime. After a hearing, the district court quashed the subpoenas on two primary grounds. First, Wyldes failed to articulate a clear connection between the requested files and his
31
case. Second, the district court determined that the information sought was not new, as any theories involving alternative suspects could have been pursued decades before his third application for postconviction relief.
We agree with the district court’s analysis. The link between the requested files and Wyldes’s current claims are attenuated at best, and to the extent that these records could have been used to identify a different perpetrator, that objective could have been achieved even before the original trial. We thus affirm the district court ruling quashing the subpoenas.
III.
For these reasons, we affirm the district court rulings that rejected each of Wyldes’s claims.
Affirmed.