Welsh v. Commonwealth: Expert Methodology Critique, Rule 2:702, and Harmless Error in Forensic Firearms Testimony

Welsh v. Commonwealth (Va. 2025): Expert Methodology Critique, Rule 2:702, and Harmless Error in Forensic Firearms Testimony

I. Introduction

The Supreme Court of Virginia’s decision in Welsh v. Commonwealth, Record No. 230800 (Mar. 20, 2025), is a major opinion at the intersection of forensic science, expert evidence, and harmless error review.

The Court:

  • Clarifies the meaning of the phrase “opines on the credibility of another witness” in Virginia Rule of Evidence 2:702(b),
  • Reaffirms that a “battle of the experts” is permitted and often necessary, even after a trial court has performed its gatekeeping function, and
  • Applies the non-constitutional harmless-error standard in a particularly rigorous way in the context of disputed forensic firearms testimony.

Defendant Brian Kuang-Ming Welsh was convicted of:

  • Two counts of first-degree murder (of drug dealer Rishi Manwani and his mother, Mala), and
  • Two counts of using a firearm in the commission of a felony.

At trial, the Commonwealth’s case centered on firearms examiner Cara McCarthy, who testified that cartridge casings from the crime scene “matched” a Browning Buck Mark .22 pistol connected to Welsh “to a practical certainty.” The defense sought to call William Tobin, a former FBI examiner and metallurgist, to critique McCarthy’s methodology and the scientific foundations of firearms/toolmark identification. The trial court barred Tobin’s testimony on two grounds:

  1. His testimony would supposedly be an impermissible opinion on another witness’s “credibility” under Rule 2:702(b); and
  2. He would be attacking “the underlying science,” which the court believed it had already definitively approved in a pretrial admissibility ruling on McCarthy.

The Court of Appeals assumed the exclusion was error but held it harmless. The Supreme Court disagreed, reversed, and remanded for further proceedings.

This commentary examines:

  • The facts and procedural posture,
  • The Court’s holdings on Rule 2:702(b) and on harmless error,
  • The precedents and authorities that shaped the decision,
  • The Court’s understanding of forensic science evidence in a “post-CSI” environment, and
  • The likely impact on future criminal litigation and expert evidence practice in Virginia.

II. Factual and Procedural Background

A. The underlying events and forensic focus

In 2017–2018, Welsh was addicted to drugs obtained from Rishi Manwani. His addiction produced severe financial distress:

  • He had depleted a joint savings account with his wife,
  • His mother had to pay off a $20,000 loan for him, and
  • He was terminated by Lending Tree on January 23, 2018, for attendance and performance issues and for keeping knives at his desk.

The HR professional who fired him, Cheri Hostetler, was sufficiently concerned about his emotional state that she requested a police welfare check. The company also issued a no-trespass notice, an atypical measure.

On January 29, 2018:

  • Carlos Rodriguez went to Rishi’s home to buy Adderall and marijuana. Welsh was outside in his car when Rodriguez arrived.
  • Rishi condescendingly referred to Welsh as his “dumb-ass friend.”
  • They then drove to a Bank of America (Rishi in Welsh’s car; Rodriguez separately). Rodriguez withdrew money and paid Rishi $200.
  • Rishi told Rodriguez not to contact him later in the week because he was going to visit his father in the hospital.

Key timing evidence:

  • 10:17 a.m.: Welsh’s phone connects to a tower near Rishi’s home.
  • 10:23 a.m.: Rishi receives a text to his “burner” phone but never reads it.
  • 10:31 a.m.: Mala last interacts with her work email.
  • 10:38 a.m.: Someone presses “2” on Rishi’s burner phone.
  • 10:39 a.m.: Welsh’s phone connects to a tower 5 miles north of Rishi’s home.
  • 10:57–10:58 a.m.: Welsh calls his mother-in-law and his mother; his phone pings near his own home.
  • 11:10 a.m.: He has a 20-minute call with a recruiter from his mother-in-law’s driveway; the recruiter testifies Welsh seemed normal and engaged.

Later that day and into the evening:

  • Welsh spends time at his mother-in-law’s house and appears normal; she does not observe blood or smell gunpowder.
  • At 12:49 p.m., he texts Rishi about a recruiting company and potential job leads.
  • Around 5:30 p.m., he asks his brother Michael to pick up a handgun from him because his wife supposedly wanted it out of the house.
  • That evening, he gives Michael a Browning Buck Mark .22 pistol in a gun safe, with Gemtech ammunition and an empty magazine.

On January 30, Welsh visits friend Caitlin McCarthy (no relation to the examiner), who knows Rishi. He appears anxious and “panicked,” says he hasn’t heard from Rishi, and suggests Rishi may owe someone over $1,000.

On January 31 and thereafter, police conduct a welfare check after Mala’s co-workers report her absence. They discover:

  • Mala dead, shot four times in the back of the head.
  • Rishi dead, shot eight times (seven to the head, one to the leg).
  • Estimated time of death: 48–72 hours prior; Rishi’s greater decomposition attributed to drugs in his system.
  • Gemtech cartridge casings near the bodies.
  • Rishi’s wallet on his body with no money; DNA matching two contributors (neither Welsh, Mala, nor Rodriguez). The “unknown” contributor eventually ruled out as the killer.
  • Single fingerprint belonging to Daniel Suh (a frequent buyer), also ruled out.
  • Rishi’s unleashed pit bull in the house, with no feces or urine and food in the bowl—suggesting the dog had been cared for during the relevant timeframe.

Police never find Rishi’s burner phone.

B. Investigation, firearms evidence, and shifting gun custody

After news of the murders, Caitlin contacts police and provides Welsh’s information. Welsh cooperates:

  • He gives fingerprints, DNA, and phone contents voluntarily.
  • He admits financial problems and drug purchases from Rishi, including draining $10,000 from his retirement fund without telling his wife.
  • He claims to have loaned Rishi $600 the day before the murder and says Rishi had about $3,000 cash when they left the bank.
  • He denies involvement in the killings and suggests other possible motives and suspects related to Rishi’s drug activities.
  • He acknowledges owning several guns but initially does not mention the Buck Mark pistol.

Police execute a search warrant at Welsh’s home and find:

  • Numerous guns,
  • .22 caliber Gemtech ammunition, and
  • Various gun parts.

After this search, Welsh has Michael return the Buck Mark so it can supposedly be given back to their father (the registered owner). Welsh then:

  • Gives the Buck Mark to his mother, and
  • Retrieves from his father a different gun registered to Welsh.

On March 20, 2018, during custodial questioning after his arrest, Welsh equivocates about ownership of the Buck Mark, then acknowledges it is his father’s gun. Police search his parents’ house and find:

  • The Buck Mark pistol,
  • Other guns,
  • Gemtech ammunition,
  • Gun parts and barrels compatible with the Buck Mark.

Three months later, in a recorded jail call, Welsh tells his father to “get rid of the soda can.” His father replies that he already “threw it out,” having “took it apart and threw it out.” Other testimony establishes that the family stored ammunition and barrels in “ammo cans.” The Commonwealth asks the jury to infer that “soda can” was code for an “ammo can” containing the barrel used in the murders.

C. The Commonwealth’s expert: Cara McCarthy

At a pretrial hearing, the prosecution describes forensic firearms examiner Cara McCarthy as “definitely the Commonwealth’s most significant witness” and “one of the most important witnesses in this case.”

McCarthy is a Department of Forensic Science firearms examiner and member of the Association of Firearm and Tool Mark Examiners (AFTE). She testifies that:

  • Firearms can leave unique microscopic markings on cartridge cases and bullets.
  • AFTE’s standard allows opinions of “common origin” when there is “sufficient agreement” in the unique surface contours of toolmarks.
  • “Sufficient agreement” means significant duplication of toolmarks such that it is “very remote” that another tool created the markings.
  • Studies support the proposition that even consecutively manufactured firearms produce distinguishable patterns.

She distinguishes:

  • Class characteristics (caliber, number and width of lands and grooves, rifling patterns, general firearm design features), from
  • Individual characteristics (random microscopic imperfections from manufacturing and use, which examiners rely on for identifications).

Her ultimate opinions at trial:

  • The cartridge cases at the Manwani residence were fired from Welsh’s Browning Buck Mark pistol.
  • She expresses this conclusion as not 100% absolute (because she has not tested every firearm in the world) but at a “practical certainty” or “very high level of certainty.”
  • The bullets that killed the victims did not come from the barrel currently on the Buck Mark; she hypothesizes that the barrel was changed, noting damaged barrel screws and mismatching bullet markings.

On cross-examination, McCarthy acknowledges:

  • Firearms/toolmark identification has no formally established error rate.
  • The final analysis is “subjective in nature,” grounded in her training and experience but “founded on scientific principles.”
  • There is no fixed numerical or statistical standard for declaring a “match”; “sufficient agreement” is a qualitative judgment.
  • Subclass characteristics (microscopic defects from the manufacturing tools themselves, shared by multiple weapons) can be a concern, but she did not evaluate the specific manufacturing tools used to make Welsh’s pistol.
  • She saw no indication that the toolmarks she relied on were mere subclass characteristics.

Defense counsel attempts to confront McCarthy with the 2016 PCAST report and the National Academy of Sciences (NAS) report, both critical of traditional pattern-matching disciplines like firearms comparison. The trial court excludes both reports as hearsay because McCarthy says the field does not rely on them. The judge permits only a limited proffer outside the jury’s presence.

D. The defense expert: William Tobin, and his exclusion

The defense then offers William Tobin as an expert in forensic metallurgy and materials science with substantial toolmark/firearms experience. In a detailed proffer, counsel represents that Tobin:

  • Worked 27 years for the FBI, conducting and supervising firearms and toolmark examinations,
  • Has conducted extensive research on forensic metallurgy, including bullet and lead analysis,
  • Has overseen a firearms manufacturing plant, giving him direct insight into manufacturing processes,
  • Has published in the field of forensic metallurgy and related disciplines,
  • Reviewed McCarthy’s file, worksheets, and testimony.

Substantively, Tobin would:

  • Explain that McCarthy’s claim that manufacturing has no impact on identifiability is scientifically incorrect.
  • Testify that an examiner cannot reliably identify a “match” without identifying and accounting for subclass characteristics, which requires investigating manufacturing processes.
  • Explain why consecutively manufactured firearms may not leave reliably distinguishable marks and why the literature does not support the AFTE “sufficient agreement” framework as a scientifically validated technique.
  • Describe how metallurgy, material science, and tribology (study of friction, wear, and lubrication) control the production of firearms and thus the toolmarks.
  • Address the absence of a known error rate and the implications for claims of “practical certainty.”
  • Critique McCarthy’s specific methods, protocols, and conclusions in this case, without opining that she was lying or acting in bad faith.

The trial court excludes Tobin on two articulated grounds:

  1. Credibility-opinion rationale: Because he reviewed McCarthy’s work and would “comment on her work,” the court deems this an impermissible opinion on her credibility under Rule 2:702(b).
  2. “Underlying science” rationale: Tobin would attack the underlying scientific basis of firearms/toolmark identification, which the court believes it already approved pretrial when admitting McCarthy’s testimony.

The judge also denies defense counsel’s request to have Tobin testify outside the jury’s presence to create a more detailed record; counsel is forced to proffer Tobin’s views by summary, which the Supreme Court later holds is sufficient for appellate review.

E. Trial outcome and appellate history

In closing, the Commonwealth heavily emphasizes McCarthy’s testimony, her training, her methodology, and her multi-stage examination of casings, bullets, and fragments. It underscores that she could have declared her results inconclusive but instead found a “match” to Welsh’s Buck Mark and concluded that the current barrel was not the one used at the crime scene. The prosecution calls attacks on her “a red herring,” asserting she did “every reasonable test” and “every necessary test.”

The defense, recognizing the centrality of her role, tells the jury the Commonwealth’s case is “built on” McCarthy’s testimony and attacks it largely through cross-examination and argument, without the benefit of Tobin’s expert counterpoint.

After several days of deliberation, the jury convicts on all counts and imposes two life sentences plus six years.

On appeal, Welsh challenges, among other things, the exclusion of Tobin under Rule 2:702 and the refusal to permit a fuller proffer. The Court of Appeals (78 Va. App. 287 (2023)):

  • Assumes, without deciding, that the exclusion was error,
  • Finds the error harmless in light of other evidence, and
  • Affirms the convictions.

The Supreme Court grants further review and reverses.

III. Summary of the Supreme Court’s Decision

The Supreme Court’s core holdings are:

  1. Rule 2:702(b) does not bar expert testimony that criticizes another expert’s methodology or conclusions. The prohibition on “opin[ing] on the credibility of another witness” refers to ad hominem or character-based veracity opinions (e.g., “she is lying,” “he is a dishonest person”), not to scientific or methodological criticism that might cause a jury to disbelieve the substance of another expert’s testimony.
  2. A pretrial ruling that an expert’s discipline is sufficiently reliable to go to the jury does not insulate that discipline from criticism by competing experts. Once the trial court has admitted one expert’s testimony, it may not bar a well-qualified opposing expert from challenging the underlying science or methodology on the theory that the court has “already ruled” on the science. Doing so invades the jury’s factfinding role.
  3. The exclusion of Tobin’s testimony was not harmless. Applying the non-constitutional harmless-error standard under Code § 8.01-678 and cases like Kilpatrick and Kotteakos, the Court finds that McCarthy’s testimony was central to the Commonwealth’s case and that denying the jury a qualified expert counterweight creates at least a “grave doubt” whether the error affected the verdict. The convictions must therefore be reversed.

Justice Kelsey, joined by Justices McCullough and Chafin, concurs that the trial court misapplied Rule 2:702 but dissents on harmless error, adopting the Court of Appeals’ view that the remaining evidence rendered the error harmless.

IV. Detailed Legal Analysis

A. Rule 2:702(b) and what it means to “opine on the credibility of another witness”

1. The text and its context

Rule 2:702(b) provides:

Expert testimony may include opinions of the witness established with a reasonable degree of probability, or it may address empirical data from which such probability may be established in the mind of the finder of fact. Testimony that is speculative, or which opines on the credibility of another witness, is not admissible.

The key interpretive question: Does an expert “opine on the credibility of another witness” whenever the expert’s testimony, if believed, would lead the jury to disbelieve another expert? Or is the Rule narrower, targeting only direct or implicit statements about another witness’s honesty or character?

The Supreme Court adopts the narrower, more traditional construction, grounded in both case law and Rule 2:102 (which directs courts to interpret the Evidence Rules consistently with pre-existing common-law principles).

2. Prior Virginia cases on credibility opinions

The Court relies on a line of cases making a clear distinction between:

  • Permissible testimony that conflicts with another witness’s version of events or another expert’s opinion, and
  • Impermissible testimony about whether that witness is truthful or capable of lying.

Key precedents include:

  • James v. Commonwealth, 254 Va. 95 (1997). The Court reiterated that credibility is for the trier of fact and held that no witness may be asked to opine whether another witness is “capable of lying.” Questions about veracity improperly invade the jury’s province.
  • Pritchett v. Commonwealth, 263 Va. 182 (2002). An expert may not express an opinion about the veracity of another witness; again, this is a jury function.
  • Fitzgerald v. Commonwealth, 223 Va. 615 (1982). The Court recognized a difference between impermissible direct commentary on veracity and permissible testimony that, by its substance, may lead a jury to disbelieve another witness without referencing that witness’s character for truthfulness.

The Court also references Rule 2:608, which provides the limited circumstances in which a witness’s character for truthfulness or untruthfulness may be attacked or supported by reputation or opinion evidence. Outside those confines, “ad hominem” credibility attacks are barred.

3. “Battle of the experts” and Grattan v. Commonwealth

The Court explicitly endorses what it calls the “proverbial ‘battle of the experts.’” When experts disagree, each may criticize the other’s methods and conclusions; the factfinder decides whom to believe.

Grattan v. Commonwealth, 278 Va. 602 (2009), is particularly important. In Grattan:

  • Competing experts criticized each other’s evaluation methods.
  • The Court did not treat those mutual critiques as impermissible “credibility” opinions.
  • Instead, it recognized that such conflict is for the jury to resolve.

The Welsh Court also cites Riner v. Commonwealth, 268 Va. 296 (2004), and Mercer v. Commonwealth, 259 Va. 235 (2000), echoing the principle that disagreement among qualified experts does not render their testimony inadmissible—it creates an issue of weight, not admissibility.

Drawing on both case law and commentary (notably Professor Kent Sinclair’s The Law of Evidence in Virginia, § 13-16), the Court underscores:

  • A challenge to an expert’s methods and determinations is not a prohibited credibility attack.
  • Disagreement among experts does not nullify their probative value.
  • The jury must determine what weight to give each expert.

4. Application to Tobin’s proffered testimony

The Court concludes that Tobin’s proffer clearly fell on the “methodology critique” side of the line:

  • He would not say that McCarthy was lying or that she was generally untruthful.
  • He would analyze whether her method was scientifically defensible, properly applied, and consistent with current empirical literature.
  • He would explain how manufacturing processes, subclass characteristics, and the lack of known error rates affected the reliability of identifications such as hers.

That such testimony, if believed, might cause the jury to disbelieve McCarthy’s conclusion does not make it a forbidden opinion on her “credibility”; it is simply the essence of adversarial expert practice.

Accordingly, the trial court’s reliance on Rule 2:702(b) to exclude Tobin was a legal error—an abuse of discretion via misapplication of the rule.

B. Expert gatekeeping and the improper notion of “science already settled by the court”

The trial court’s second rationale—that Tobin was improperly “attacking the underlying science” which the court had already “ruled on” pretrial—also fails.

1. The trial court’s gatekeeping role

Virginia trial courts, like their federal counterparts, act as evidentiary gatekeepers for expert testimony. They decide:

  • Whether the expert is qualified, and
  • Whether the expert’s field and methods have enough scientific grounding to be presented to the jury at all.

In this sense, the judge filters out “junk science.” But the Supreme Court makes a critical point: a pretrial decision that one expert’s methodology is admissible does not:

  • Transform that methodology into unquestionable “holy writ,” and
  • Does not preclude another qualified expert from testifying that the methodology is flawed, unreliable, or not generally accepted.

To the contrary, once the court determines that a field and method have passed the threshold for admissibility, the adversary process—cross-examination, competing experts, and argument—must be allowed to test its weight and persuasive force.

2. The Court’s criticism of the trial judge’s approach

The Supreme Court expressly rejects the notion that the trial court’s gatekeeping ruling insulated McCarthy from methodological challenge:

  • The judge may decide that firearm/toolmark identification meets minimum standards of reliability.
  • But that determination does not mean that another expert cannot argue that:
    • the discipline is overstated,
    • error rates are unknown,
    • critical literature (e.g., PCAST, NAS) undermines the claimed certainty, or
    • the expert misapplied her own discipline’s protocols.
  • By excluding Tobin on the ground that the court had already “ruled on the science,” the trial court improperly usurped the jury’s function—to weigh conflicting expert opinions.

The Welsh Court, again drawing on Professor Sinclair, notes that when a judge discards parts of a defense expert’s testimony because they conflict with the prosecution expert’s, the judge is effectively weighing evidence and assessing credibility—functions reserved to the jury in a jury trial.

C. Harmless error: statutory framework and standards

1. Code § 8.01-678 and mandatory harmless-error review

Under Code § 8.01-678:

When it plainly appears from the record and the evidence given at the trial that the parties have had a fair trial on the merits and substantial justice has been reached, no judgment shall be arrested or reversed… for any error committed on the trial.

The Supreme Court has repeatedly emphasized that this statute:

  • Makes harmless-error review mandatory in all cases (Commonwealth v. Swann, 290 Va. 194 (2015); Ferguson v. Commonwealth, 240 Va. ix (1990)).
  • Limits the Court’s power to reverse; it may do so only where an error was not harmless (Commonwealth v. White, 293 Va. 411 (2017); Walker v. Commonwealth, 144 Va. 648 (1926)).

2. Constitutional vs. non-constitutional error

The Court distinguishes:

  • Constitutional errors (e.g., violations of the Confrontation Clause), governed by the familiar Chapman v. California, 386 U.S. 18 (1967), standard: An appellate court must be convinced beyond a reasonable doubt that the error did not contribute to the verdict.
  • Non-constitutional errors (as here, involving application of evidentiary rules), where the test is whether:
    • “the error did not influence the jury or had but slight effect” (Commonwealth v. Kilpatrick, 301 Va. 214 (2022), citing Kotteakos v. United States, 328 U.S. 750 (1946)), and
    • the evidence of guilt is “so overwhelming” that the error could not have affected the outcome (Haas v. Commonwealth, 299 Va. 465 (2021)).

The Court also quotes Clay v. Commonwealth, 262 Va. 253 (2001), which, relying on Kotteakos, holds that if the reviewing court has a “grave doubt” whether the error did or did not affect the verdict, the conviction cannot stand.

3. “Standards” vs. “rules” and the case-specific nature of harmless-error review

The Court spends unusual time explaining why harmless-error doctrine is inherently fact-specific:

  • It is built on standards, not rules. A “rule” sets determinate consequences for specific factual predicates; a “standard” (like “harmless beyond a reasonable doubt” or “no substantial influence on the verdict”) requires judgment and context.
  • Citing academic commentary (Michael Coenen and Justice Scalia) and Shinseki v. Sanders, 556 U.S. 396 (2009), the Court warns against mechanical presumptions or bright-line rules in harmless-error analysis.
  • There is no trial court harmless-error ruling to which an appellate court can defer; harmlessness is assessed de novo by the appellate court based on the cold record.
  • The critical question is not whether the appellate judges themselves, in hindsight, would have convicted absent the error; the question is what effect the error likely had on the jury that actually decided the case (Kotteakos).

The Court acknowledges that reasonable and conscientious judges can disagree about whether a given error was harmless, echoing its prior observation in Minh Duy Du v. Commonwealth, 292 Va. 555 (2016), that the same facts can reasonably support different legal conclusions.

D. Applying the harmless-error standard in Welsh

1. The nature of the error: what Welsh was denied

The Court first asks: what did the error actually deprive the defendant of?

Answer: the opportunity to present a qualified, highly experienced forensic expert (Tobin) to:

  • Directly challenge the scientific foundation of the key forensic discipline used against him (firearms/toolmark identification), and
  • Specifically critique the methods and strength of the Commonwealth’s central expert’s conclusions tying his gun to the casings.

The Court explicitly recognizes the power of scientific evidence in a “post-CSI world.” It notes the risk described in United States v. Green, 405 F. Supp. 2d 104 (D. Mass. 2005): jurors may give forensic testimony labeled “scientific” “more credence than it deserves.” That risk is magnified where:

  • The jury hears only one side of the scientific debate, and
  • The trial court erroneously excludes a credible counter-expert.

Because the proffer shows that Tobin was well qualified on these issues, the Court cannot say that reasonable jurors would not have found his criticisms persuasive enough to significantly discount, or even reject, McCarthy’s “practical certainty” that Welsh’s gun was the source of the casings.

2. The centrality of McCarthy’s testimony

The Court next examines how important McCarthy’s evidence was to the prosecution’s case:

  • There were no eyewitnesses to the murders and no video evidence.
  • The case was circumstantial and inference-driven.
  • McCarthy’s testimony provided a direct forensic link between Welsh’s family’s firearm and the murder casings—bridging the gap from opportunity and motive to physical connection.

The Commonwealth itself repeatedly characterized her as:

  • “the Commonwealth’s most significant witness,”
  • “one of the most important witnesses in this case.”

And in closing arguments, the prosecution:

  • Devoted extensive time to explaining and emphasizing her methodology and findings,
  • Argued she had performed “every reasonable test” and swung back at defense critique as a mere “red herring,”
  • Relied on her testimony to underpin its argument that Welsh’s family had reason to hide or destroy the relevant barrel.

The defense, for its part, told the jury that the Commonwealth’s case was “built on” her testimony.

3. The other evidence of guilt

The Court acknowledges that, even without McCarthy, the Commonwealth introduced substantial inculpatory evidence. It summarizes the other key evidence:

  • Welsh’s presence at Rishi’s home shortly before the estimated time of death.
  • His financial desperation and substantial drug-related debt.
  • His inconsistent statements to police about his gun ownership and his timeline after leaving Rishi.
  • His possession and movement of the Buck Mark around family members, including the timing of these transfers around the search and investigation.
  • The recorded instruction to his father to “get rid of the soda can” and the father’s admission that he had disassembled and discarded it, in the context of “ammo cans” used to store barrels and ammunition.
  • His attempts to manage how his relationships with witnesses like Michael and Caitlin appeared to investigators.

If the only question were sufficiency of the evidence without McCarthy’s testimony, the Court indicates that the answer would likely be that the evidence was sufficient for conviction. But sufficiency and harmlessness are not the same inquiry.

4. Why the Court has “grave doubt” and cannot find harmlessness

The central question under the non-constitutional harmless-error standard: did the exclusion of Tobin’s testimony “not influence the jury or [have] but slight effect”? Or does the Court have a “grave doubt” whether it affected the verdict?

The Court answers the latter. Several factors together create this grave doubt:

  • Nature of the evidence excluded: This was not marginal or cumulative. Tobin’s testimony would have gone directly to the reliability of the linchpin forensic identification tying Welsh’s gun to the casings.
  • Nature of the evidence admitted: McCarthy’s testimony was highly technical, presented as expert forensic science, and repeatedly framed as central by the Commonwealth.
  • Asymmetry in presentation: The jury heard a full, confident exposition of AFTE methodology and the concept of “practical certainty,” but was denied the countervailing scientific critique that would have contextualized and potentially undercut these claims.
  • Susceptibility of jurors to scientific evidence: The Court expressly notes the risk in a “post-CSI” media environment that jurors will overvalue such evidence.
  • Presence of some defense evidence suggesting alternative scenarios: –
    • Evidence that Rishi’s drug supply was unstable and he had conflicts or debts,
    • Testimony that Rishi appeared bruised and nervous in the period before his death, possibly indicating other threats,
    • The neighbor’s testimony (albeit equivocal) about seeing Rishi’s pit bull outside with a tall person the day the Commonwealth claimed the victims were already dead.
  • Case-specific nature of harmless-error analysis: The Court stresses that its conclusion flows from the particular constellation of facts in Welsh, not from a rigid rule that all exclusions of defense forensic experts are per se prejudicial.

In this context, the Court cannot say the error “did not influence the jury or had but slight effect.” It has at least a “grave doubt” whether, had jurors heard Tobin’s critique:

  • They might have viewed the firearm/toolmark identification discipline more skeptically,
  • Given less weight to McCarthy’s “practical certainty,” and
  • Possibly harbored a reasonable doubt about whether Welsh used that specific firearm to commit the murders.

Because such grave doubt exists, the error is not harmless; reversal is required.

5. The concurring and dissenting view

Justice Kelsey (joined by Justices McCullough and Chafin) agrees with the majority’s Rule 2:702 analysis but believes the error harmless. He would adopt the Court of Appeals’ reasoning that the non-forensic evidence of guilt was sufficiently strong that Tobin’s testimony would not likely have changed the result.

This split underscores the fact-intensive, judgment-laden nature of harmless-error review: reasonable jurists assessing the same record can differ about the likely impact of the error on the jury.

V. Precedents and Authorities Shaping the Opinion

Beyond the core cases already discussed, the Court draws on a substantial body of law:

  • Midkiff v. Commonwealth, 280 Va. 216 (2010); Carter v. Commonwealth, 293 Va. 537 (2017); Lawrence v. Commonwealth, 279 Va. 490 (2010). These cases define the abuse-of-discretion standard for evidentiary rulings and the idea that misapplication of legal principles is itself an abuse of discretion.
  • Findlay v. Commonwealth, 287 Va. 111 (2014); LaCava v. Commonwealth, 283 Va. 465 (2012). They confirm that interpretation of the Rules of Evidence presents a pure question of law, reviewed de novo.
  • Wyche v. Commonwealth, 218 Va. 839 (1978); Whittaker v. Commonwealth, 217 Va. 966 (1977); Watkins v. Commonwealth, 229 Va. 469 (1985); Peterson v. Castano, 260 Va. 299 (2000). These cases address proffers of excluded testimony. The Court notes that although a proffer may be made by counsel’s avowal, “the better practice” is to let the witness testify outside the jury’s presence. The Court does not decide whether the refusal to permit Tobin to testify outside the jury was reversible error because the existing proffer sufficed.
  • Commonwealth v. Swann, 290 Va. 194 (2015); Commonwealth v. White, 293 Va. 411 (2017); Ferguson v. Commonwealth, 240 Va. ix (1990); Walker v. Commonwealth, 144 Va. 648 (1926). These cases interpret Code § 8.01-678 and emphasize that harmless-error review is mandatory and limits appellate authority to reverse.
  • Commonwealth v. Kilpatrick, 301 Va. 214 (2022); Haas v. Commonwealth, 299 Va. 465 (2021); Salahuddin v. Commonwealth, 67 Va. App. 190 (2017). These decisions refine the non-constitutional harmless-error test and the “overwhelming evidence” concept.
  • Chapman v. California, 386 U.S. 18 (1967); Kotteakos v. United States, 328 U.S. 750 (1946); Shinseki v. Sanders, 556 U.S. 396 (2009). The U.S. Supreme Court authorities that provide the doctrinal backbone for harmless-error standards and caution against rigid, presumptive approaches.
  • Saal v. Commonwealth, 72 Va. App. 413 (2020); Wynnycky v. Kozel, 71 Va. App. 177 (2019). The Court of Appeals’ observation that “legal standards” are not readily reducible to hard-and-fast “rules,” which the Supreme Court embraces.
  • Minh Duy Du v. Commonwealth, 292 Va. 555 (2016); Thomas v. Commonwealth, 62 Va. App. 104 (2013). Cited for the proposition that different reasonable jurists may reach different conclusions based on the same facts without either being unreasonable.
  • California v. Green, 399 U.S. 149 (1970); Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579 (1993). Green for its characterization of cross-examination as the “greatest legal engine” for discovering truth; Daubert as recognizing that cross-examination and competing evidence (including expert evidence) are traditional tools for attacking “shaky but admissible” expert testimony.

VI. Complex Concepts and Terminology Explained

A. Firearms and toolmark identification

Firearms examiners like McCarthy and (historically) Tobin work in the field of toolmark identification, which includes:

  • Comparing bullets and cartridge cases recovered from crime scenes to test-fired exemplars from suspect firearms, and
  • Looking for microscopic similarities in toolmarks produced by components like firing pins, breechfaces, extractors, ejectors, and rifled barrels.

Key terms:

  • Class characteristics: Design features shared by all firearms of a type or model (e.g., caliber, number and width of barrel lands and grooves, rifling type, position of firing pin). These can narrow down a firearm category but cannot uniquely identify a single gun.
  • Individual characteristics: Microscopic irregularities from manufacturing and subsequent use (wear, corrosion). Examiners assert that these are sufficiently unique that, in combination, they can tie a particular bullet or casing to a particular firearm.
  • Subclass characteristics: Microscopic features originating from the manufacturing tools themselves (e.g., cutting tools, rifling machines). These features can be replicated across multiple firearms until the tool is replaced or extensively worn. If misinterpreted as individual characteristics, they can lead to mistaken identifications.

The AFTE standard (used by McCarthy) permits an “identification” when there is “sufficient agreement” in the comparison microscope patterns. Importantly:

  • “Sufficient agreement” has no fixed numeric definition or universally accepted statistical threshold.
  • The field lacks universally accepted error rates, although some studies exist.
  • Examiners rely on training and experience to make qualitative judgments—a point both McCarthy and Tobin acknowledge as central to the methodological debate.

B. PCAST and NAS reports

Two major external critiques of pattern-matching forensic disciplines:

  • PCAST (President’s Council of Advisors on Science and Technology) report (2016) PCAST concluded that many forensic pattern-matching techniques—including firearms/toolmark identification—did not yet meet standards of “foundational validity” and that only limited, methodologically sound studies existed.
  • National Academy of Sciences (NAS) report (2009) The NAS criticized the scientific underpinnings of several forensic disciplines and called for more rigorous research, standardized protocols, and statistical development.

In Welsh, defense counsel attempted to confront McCarthy with these reports to show that:

  • The discipline she relied on is scientifically contested,
  • Its foundational validity and error rates are in doubt, and
  • The field has been criticized for overstating certainty.

The trial court excluded these reports as hearsay because McCarthy testified the field does not rely on them. The Supreme Court does not directly resolve the admissibility of such reports via cross-examination here, but their presence underscores the wider scientific controversy into which Tobin’s testimony would have fit.

C. Proffers of excluded testimony

When a trial court excludes proposed evidence, the proponent must make a proffer (offer of proof) to preserve the issue for appeal. A proffer can be:

  • A narrative by counsel summarizing what the witness would say, or
  • The witness actually testifying outside the jury’s presence.

The Supreme Court reinforces that:

  • A proffer is essential; without it, an appellate court cannot determine whether exclusion was prejudicial (Wyche).
  • Counsel’s narrative may suffice, but allowing the witness to testify in camera is “the better practice.”
  • Efficiency cannot override a party’s right to a fair trial (Peterson v. Castano).

In Welsh, the Court finds counsel’s proffer adequate to analyze the issue and thus sidesteps whether the refusal to permit Tobin to testify outside the jury was itself reversible error. Nonetheless, the opinion strongly signals that trial courts should err on the side of permitting a live proffer where the testimony is technical and complex.

D. Harmless error vs. sufficiency of the evidence

A crucial conceptual distinction:

  • Sufficiency of the evidence asks: could a rational jury, viewing the (properly admitted) evidence in the light most favorable to the Commonwealth, find guilt beyond a reasonable doubt?
  • Harmless error asks: assuming an error occurred (e.g., wrongly excluding defense evidence), is there a reasonable probability that the error affected the actual verdict reached by the particular jury in this case?

Thus, even where the evidence (excluding the tainted or missing piece) would be legally sufficient to convict, the exclusion of important countervailing evidence can still be prejudicial and require reversal if it might have changed the jury’s mind.

VII. Practical Implications and Future Impact

A. For criminal defense counsel

Welsh meaningfully strengthens the defense’s ability to present:

  • Independent expert critiques of forensic disciplines (such as firearms/toolmark identification, fingerprints, etc.), and
  • Case-specific methodological critiques of the Commonwealth’s forensic examiners.

Key takeaways:

  • Defense experts may explicitly challenge:
    • The existence or adequacy of validation studies,
    • The lack of known error rates,
    • The AFTE “sufficient agreement” standard as subjective or non-statistical,
    • The failure to account for subclass characteristics or manufacturing variation.
  • So long as the expert does not label the opposing expert a liar or impugn her character, such testimony is not an impermissible opinion on “credibility” under Rule 2:702(b).
  • When a trial court has already ruled that a forensic discipline is “reliable enough” for admission, defense counsel should be prepared to argue—citing Welsh—that this ruling does not preclude a defense expert from criticizing that very discipline.
  • Counsel should insist on a detailed proffer (ideally through live testimony outside the jury’s presence) if a defense expert is challenged, to preserve both admissibility and harmless-error issues on appeal.

B. For prosecutors

Welsh signals that:

  • The Commonwealth’s forensic science witnesses cannot be effectively insulated from methodological challenge by invoking Rule 2:702(b) or by pointing to an earlier gatekeeping ruling.
  • Prosecutors should be prepared for true “battles of the experts” on firearms/toolmarks and other pattern-matching fields.
  • Overstating forensic certainty (“practical certainty,” “match” with no error rate) will be increasingly scrutinized—especially when foundational literature (PCAST, NAS) has been critical.
  • Strategically, prosecutors may wish to:
    • Present their experts as open to, and aware of, the limits and criticisms of their discipline, and
    • Prepare their experts for cross-examination not only on their specific steps in the case but also on the broader scientific debates.

C. For trial judges

Welsh contains explicit guidance:

  • Do not conflate threshold admissibility with unassailability. A ruling that one expert’s science is “reliable enough” to reach the jury does not authorize excluding a competing expert’s scientific critique.
  • Apply Rule 2:702(b) narrowly. The “credibility” prohibition bars ad hominem attacks on character for truthfulness, not methodological disagreement or contrary conclusions.
  • Facilitate robust proffers. When excluding expert testimony, particularly technical testimony, it is “better practice” to allow the expert to testify outside the jury’s presence to create a complete record.
  • Be mindful of harmless error. Excluding a defense expert on a central forensic issue will rarely, if ever, be harmless where the prosecution’s case heavily relies on the very forensic evidence that the expert would have challenged.

D. For the development of forensic science litigation in Virginia

Welsh positions the Supreme Court of Virginia as attentive to:

  • The national scientific debate about the reliability and error rates of traditional pattern-matching forensics, and
  • The risk of overstated certainty in examiner testimony.

While the Court does not rule on the ultimate scientific validity of AFTE-based firearm identification, it insists that juries in Virginia must be allowed to hear qualified criticisms of its foundations and applications. That principle will likely:

  • Encourage more defense challenges to firearms identifications and other pattern-matching disciplines (e.g., toolmarks, latent prints, some types of forensic comparison), and
  • Shape how Virginia courts handle Daubert-like challenges and admissibility hearings in forensic cases going forward.

E. For appellate practice and harmless-error doctrine

From an appellate perspective, Welsh:

  • Reaffirms that harmless-error review is not a mechanistic “overwhelming evidence = harmless” test.
  • Stresses the importance of evaluating:
    • The centrality of the affected evidence,
    • How the parties framed and argued it to the jury,
    • Whether other evidence served as a functional substitute, and
    • Whether, in context, the reviewing court harbors “grave doubt” about the verdict’s independence from the error.
  • Illustrates the difficulty of harmless-error analysis in cases involving disputed forensic science and competing inferences from circumstantial evidence.

VIII. Conclusion

Welsh v. Commonwealth is a significant precedent in Virginia evidence and criminal law. Its main contributions are:

  1. Clarification of Rule 2:702(b): Expert testimony that critiques another expert’s methodology, reasoning, or conclusions—however powerful its effect on the jury’s assessment—is not an impermissible opinion on “credibility” so long as it avoids ad hominem commentary on character or truthfulness. This preserves the essential “battle of the experts” in cases where scientific evidence is contested.
  2. Reaffirmation of the jury’s role: Once a trial court has admitted expert testimony, it may not shield that testimony from scientific challenge by invoking its gatekeeping ruling. Conflicts between qualified experts are for the jury to resolve.
  3. Robust harmless-error review in forensic cases: Excluding a defense expert who would directly confront the core forensic identification evidence—particularly in a circumstantial case where that forensic link is central—will rarely be harmless. The Court insists on a searching, case-specific application of the non-constitutional harmless-error standard and applies the “grave doubt” formulation with real bite.

In a legal landscape where forensic science plays an increasingly prominent role, Welsh underscores that:

  • Forensic evidence must be subjected to full adversarial testing,
  • Trial courts must carefully distinguish between admissibility and weight, and
  • Appellate courts must rigorously evaluate whether erroneous restrictions on defense evidence may have influenced the jury’s verdict.

By reversing the convictions and remanding for further proceedings, the Supreme Court of Virginia affirms a core principle: when the Commonwealth builds its case on a single, complex forensic identification, the defense must be allowed a meaningful opportunity to challenge that evidence with qualified expert testimony. Anything less impermissibly distorts the jury’s fact-finding function and risks undermining the fairness of the criminal trial.

Case Details

Year: 2025
Court: Supreme Court of Virginia

Comments