PRESENT: All the Justices
BRIAN KUANG-MING WELSH OPINION BY v. Record No. 230800 JUSTICE WESLEY G. RUSSELL, JR. MARCH 20, 2025 COMMONWEALTH OF VIRGINIA
FROM THE COURT OF APPEALS OF VIRGINIA
Brian Kuang-Ming Welsh appeals a decision of the Court of Appeals affirming his
convictions for two counts of first-degree murder and two counts of using a firearm in the
commission of a felony. Welsh asserts that the Court of Appeals erred in concluding that the
trial court’s refusal to permit testimony from his firearms expert was, at most, harmless error.
For the reasons that follow, we conclude that the trial court erred in prohibiting the testimony of
Welsh’s expert and that such error was not harmless. Accordingly, we reverse the judgment of
the Court of Appeals.
I. BACKGROUND
In 2017 and 2018, Welsh was a drug addict. He regularly obtained drugs from Rishi
Manwani (“Rishi”).
Welsh’s drug use was causing him financial problems. He had depleted a savings
account that he shared with his wife and needed his mother to pay off a $20,000 loan for him.
Welsh was distraught and did not know how to break the bad financial news to his wife.
To make matters worse, Welsh was terminated by his employer, Lending Tree, on
January 23, 2018, for “attendance [issues], falling asleep at his desk, being unprofessional,
behavior [issues,] and [having] knives at his desk.” Cheri Hostetler, the human resources
professional who fired Welsh, reported that Welsh was beside himself when she fired him.
Hostetler told Welsh that she was “worried about [him,]” and Welsh responded that “[she] should be worried about [him].” Hostetler asked the police to do a welfare check on Welsh
because she was concerned about his well-being. Later that day, Welsh called her back, thanked
her for calling the police, and asked about scheduling an exit interview. Hostetler noted that
Lending Tree issued a no trespass notice to Welsh, which was not standard procedure for a
normal termination.
On the morning of January 29, 2018, Carlos Rodriguez met with Rishi at Rishi’s house to
buy Adderall and marijuana. Rishi lived with his mother, Mala Manwani (“Mala”). Welsh was
outside Rishi’s house in his car when Rodriguez arrived. Rodriguez asked Rishi who was in the
car, and Rishi responded that it was his “dumb-ass friend.” Rishi talked condescendingly
towards Welsh once Welsh joined them in the basement. Soon thereafter, the trio left for Bank
of America so that Rodriguez could withdraw money to pay Rishi for drugs. Rishi and Welsh
rode in Welsh’s car, and Rodriguez drove separately. After the bank opened at 10 a.m.,
Rodriguez obtained money and paid Rishi $200. Before leaving, Rodriguez asked Rishi if he
wanted to get together later that week, but Rishi said not to contact him because he would be
visiting his father in the hospital.
At 10:17 a.m., Welsh’s phone connected to a cell phone tower within range of Rishi’s
home. At 10:23 a.m., Rishi received a text message on his burner phone but never read it. At
10:31 a.m., Mala, who was working from home, last interacted with her email inbox. At 10:38
a.m., somebody pressed the number “2” on Rishi’s burner phone. At 10:39 a.m., Welsh’s phone
was moving and connected to a tower five miles north of Rishi’s home. At 10:57 a.m., Welsh
called his mother-in-law and told her that he was coming to her house to pick up his kids. He
called his mother a minute later and told her the same thing. His phone connected to a cell phone
tower near his home during both of these calls. At 11:10 a.m., Welsh spoke to a recruiter from
2 an IT company for 20 minutes. The recruiter testified that Welsh seemed normal during the call,
that it went well, and that he sought to move forward with Welsh as an applicant. Welsh spoke
to the recruiter while in his mother-in-law’s driveway. After the call, he went inside his mother-
in-law’s house and hugged her. His mother-in-law testified that he seemed normal and that she
did not smell gunpowder or see any blood on him.
At 12:40 p.m., Welsh texted his brother, Michael, and asked Michael to call when he
could. At 12:43 p.m., Welsh spoke to his mother on the phone again, and this time connected to
a cell tower near his home. At 12:49 p.m., he texted Rishi, saying “[h]ey, call me when you can
if you aren’t in the hospital. A recruiting company called Kforce called me this morning with a
job offer in D.C. but they seem to have a bunch of data center openings that you’d be a good fit
for.” Welsh never tried contacting Rishi again. Michael responded to Welsh at 1:23 p.m.,
saying that he could not call him until after he got off of work and asking him “what[ was] up?”
Welsh texted him back at 1:28 p.m. and told him to “[j]ust call me after work.” Michael called
Welsh around 5:30 p.m. During that call, Welsh asked Michael to pick up a handgun from him,
claiming that his wife wanted it out of the house. That evening, Welsh gave Michael the
handgun, a Browning Buck Mark .22 pistol, in a gun safe along with Gemtech ammunition and
an empty magazine.
On January 30, 2018, Welsh drove to the house of his friend, Caitlin McCarthy
(“Caitlin”). He asked Caitlin, who was also a friend of Rishi’s, if she had heard from Rishi.
Welsh indicated that he was concerned about Rishi because he had not heard from him since the
day before and both of Rishi’s phones were off. Welsh said that Rishi normally got back to him
within an hour or two. Caitlin also had noticed that Rishi’s phones were off when she tried to
call him but did not think much of it at the time. Caitlin testified that Welsh seemed panicked
3 and anxious during their talk. Welsh then asked if Caitlin knew of anybody that may have
wanted to hurt Rishi, explaining that he believed Rishi owed somebody over a thousand dollars.
Mala’s coworkers became concerned about her when she did not come into work on
Tuesday, January 30. None of them had heard from her since Monday morning. When Mala did
not show up on Wednesday, January 31, they called the police to conduct a welfare check.
Thereafter, the police discovered that Mala and Rishi were dead. Mala had suffered four gunshot
wounds to the back of the head. Rishi had suffered seven gunshot wounds to the head and one to
the leg. The police determined that they likely had been dead for between 48 and 72 hours.
Rishi’s body was significantly more decomposed than Mala’s, which the medical examiner
determined “[was] due to the presence of drugs in [Rishi’s] system.” Gemtech cartridge casings
were found near the bodies.
The police never found Rishi’s burner phone. However, Rishi’s wallet was found on top
of his body with no money in it. The DNA of Rishi and one other person was found on the
wallet—Rodriguez, Welsh, and Mala were all eliminated as contributors. The Commonwealth’s
DNA analyst uploaded the genetic profile into a database and found five potential matches, four
of which he eliminated and one of which he could not eliminate. However, the police
determined that this person did not commit the murders.
The police also found a fingerprint of Daniel Suh, one of Rishi’s buyers who frequented
Rishi’s house, on the backdoor leading into the basement where Rishi lived, but they also
determined that he did not commit the murders. The police did not find any other fingerprints
that did not belong to either Mala or Rishi.
4 Finally, Rishi had a dog, a white pit bull, that was unleashed in the house when the police
arrived. The police did not discover any dog feces or urine in the house, and the dog had food in
its bowl.
Caitlin texted Welsh on February 1, 2018, after hearing about Rishi’s murder. She told
him that she should have listened to him and that he was right to be concerned about Rishi’s
safety. The two spoke the following day and tried to determine who would want to hurt Rishi.
Welsh deleted his messages and communications with Caitlin from around this time period.
Caitlin was not surprised about that because Welsh’s wife, Amy, did not approve of their
relationship. Caitlin reached out to and talked to the police shortly after she spoke with Welsh.
During her discussion with the police, Caitlin gave them Welsh’s contact information.
The police interviewed Welsh numerous times over the following weeks. Welsh
voluntarily provided his fingerprints, DNA, and the contents of his phone. Welsh told the police
about his financial troubles and how he drained $10,000 from his retirement account to buy
drugs from Rishi without telling his wife. He also claimed to have loaned Rishi $600 the day
before Rishi was killed. Welsh stated that Rishi had about $3,000 in cash in his wallet when they
left the bank on January 29.
Welsh denied killing Rishi and Mala, explaining that he promptly left Rishi’s house after
returning from the bank because Rishi planned to go visit his father in the hospital. Welsh
hypothesized that Rishi’s death was linked to him needing money and offered several leads on
who may have wanted to hurt Rishi. Welsh also admitted to owning several guns but did not
mention the Buck Mark pistol.
After a few interviews, the police executed a search warrant at Welsh’s home and found
numerous guns, .22 caliber Gemtech ammunition, and various gun parts. Shortly after this
5 search, Welsh asked Michael to return the Buck Mark pistol so that he could give it to their
father, who was the registered owner. Welsh explained that he wanted all of the family’s
weapons to be returned to their registered owners. Michael acknowledged that Welsh had given
him guns before because Amy wanted them out of the house. Michael returned the Buck Mark
to Welsh. Welsh later gave his mother the Buck Mark pistol and reclaimed a gun his father had
that was registered to Welsh.
Welsh was arrested and questioned on March 20, 2018. Welsh denied that the Buck
Mark pistol belonged to him, but eventually stated that it was his father’s gun. The police
searched Welsh’s parents’ house and recovered the Buck Mark pistol. The police also recovered
other guns, Gemtech ammunition, gun parts, and gun barrels compatible with the Buck Mark.
Three months after his arrest, Welsh told his father on a recorded call from jail to “[g]et
rid of the soda can.” His father replied, “[y]eah, I did. I threw it out. I took it apart and threw it
out.” Other evidence established that Welsh’s family stored ammunition and gun barrels in
“ammo cans.” The Commonwealth argues that, by inference, a reasonable factfinder could
conclude that Welsh’s reference to throwing away a soda can was coded language for disposing
of an ammunition can that concealed the gun barrel used in the murders.
One of the central aspects of the Commonwealth’s case against Welsh was the testimony
of Cara McCarthy, “the Department of Forensic Science firearms examiner[.]” At a pretrial
hearing, the Commonwealth described McCarthy as “definitely the Commonwealth’s most
significant witness[.]” The Commonwealth reiterated the significance of McCarthy’s testimony
to its case, stating that “one of the most important witnesses in this case is Cara McCarthy.”
Based on her forensic examination, McCarthy concluded that Welsh’s Buck Mark .22
pistol “matched” the cartridge casings found in the Manwani residence, but that the barrel
6 attached to the gun at the time it was seized did not match the bullets found in the residence. She
posited that the barrel may have been switched out because the screws on the barrel appeared
damaged.
Before trial, Welsh moved to preclude McCarthy from testifying, alleging that the
scientific foundation underlying her opinion was unsound. McCarthy testified pre-trial that she
was a member of the Firearm and Toolmark Examiner’s Association (AFTE). She stated that in
order to determine whether a firearm “matches” a cartridge case, firearm testing must reveal that
the firearm left unique markings on that cartridge case. She testified that multiple studies had
found that even consecutively manufactured firearms leave unique and identifiable marks on
cartridge cases when the gun is fired. Welsh claimed otherwise. He asked McCarthy about the
President’s Council of Advisors of Science and Technology (PCAST) report claiming that there
had been only one valid study testing the scientific basis of the practice. McCarthy testified that
that report failed to recognize other valid studies that had been done and that PCAST’s criteria
for what constituted a “valid” study were unduly strict. She also claimed that various
organizations and agencies had discredited the report due to statistical errors and failing to
recognize other valid studies. She contended that the report was not relied upon in her industry.
McCarthy conceded that her “analysis is subjective in nature” and that “this is due to the
evaluation of those individual characteristics,” i.e., the unique characteristics of the firearm that
imprint on the cartridge case, “and the significance of those individual characteristics.” She
explained that there is no set scientific standard to determine whether a firearm “matches” a
cartridge case. She stated that examiners use the standard set forth by the AFTE, which “allows
opinions of common origin to be made when the unique surface contours of toolmarks are in
sufficient agreement.” She testified that under the AFTE’s standard, two samples are in
7 “sufficient agreement” when there is “significant duplication of toolmark[s] evidenced by
correspondence of patterns, a combination of patterns, because the quality and quantity of the
likelihood that another tool created the markings is very remote.”
The trial court determined that the Commonwealth had demonstrated that McCarthy’s
opinions were sufficiently reliable to be considered by the jury and overruled Welsh’s motion to
prohibit McCarthy from testifying.
The Commonwealth called McCarthy to testify at trial. She testified that she evaluated
the Browning Buck Mark .22 pistol and the cartridge cases found at the Manwani’s for class
characteristics and individual characteristics. She explained that class characteristics “are
features that are determined prior to manufacture by the manufacturer. Class characteristics can
include caliber, the number of lands and grooves, the width of the lands and grooves. Lands and
grooves make up the rifling of the barrel, rifling to impart a spin on the projectile for stability
and accuracy, the location of the extractor and the ejector in the firearm, among other general
characteristics.” She contrasted class characteristics with individual characteristics, which “are
random microscopic imperfections that are created during the manufacturing process. They’re
not intended to be there by the manufacturer. And these are what we use to make opinions of
identification if a bullet or a cartridge case was fired from a particular firearm.”
McCarthy then stated her opinion that Welsh’s Buck Mark pistol matched the cartridge
cases found at the Manwani’s. She stated that her opinion is “not 100-percent absolute certainty”
because she “[has] not examined every single firearm in the world.” However, she followed that
qualification by saying that her opinion was given to “a practical certainty or very high level of
certainty.” McCarthy then claimed that the bullets that killed Rishi and Mala were fired from a
different barrel than the one then attached to the Buck Mark, pointing to the damaged screws on
8 the barrel of the firearm and the mismatch of the individual characteristics she found when
comparing the bullets to the barrel.
Under cross-examination, McCarthy stated that firearm and toolmark identification has
no established error rate. McCarthy conceded that “[t]he final analysis and conclusion is
subjective in nature. It’s based on [her] training and experience, but it is founded on scientific
principles.” She testified that she may find “sufficient agreement” between a firearm and a
cartridge case when their patterns “exceed[] any agreement or even the best agreement that
would be exhibited between two firing pin impressions that I know came from two different
firearms, but it also has to be consistent in the quality and quantity of the agreement that I would
expect to see between two firing pin impressions that I know came from the same firearm.”
Welsh then pressed McCarthy to explain how she factors subclass characteristics into her
analysis. Subclass characteristics, she stated, are essentially miniscule defects that arise via
random imperfections on the equipment used to manufacture the firearm. These imperfections
would be present on all firearms created by that piece of manufacturing equipment until it is
altered by the manufacturer, or the equipment gets worn down.
McCarthy acknowledged that she did not examine the equipment used to manufacture
Welsh’s Buck Mark or study its manufacturing process before rendering her opinions. McCarthy
conceded that she could not “make an opinion of identification on subclass characteristics,” but
also stated that examiners like her are trained to recognize subclass characteristics and that
“[t]here was no indication during [her] examination that the tool marks that [were] within the
firing pin impression on those cartridge cases were subclass characteristics.” She maintained
that trained examiners like herself can distinguish between consecutively manufactured firearms.
She noted that consecutively manufactured firearms may not have consecutively manufactured
9 firing pins, and so the mark left by a firing pin may not share a subclass characteristic with a
consecutively manufactured firearm.
McCarthy acknowledged that a gun’s barrel may be removed to clean the firearm, and
that she could not tell whether the screws on the barrel in this case were original, how many
times they were removed, or when they were taken out.
Welsh sought to cross-examine McCarthy using the PCAST report he referenced pre-
trial, but the trial court denied its introduction on hearsay grounds. The trial court permitted
Welsh to make a proffer by questioning McCarthy outside of the presence of the jury. During
that proffer, McCarthy contended that people in her profession did not rely on the report and that
it had been rejected by numerous organizations and agencies, such as the International
Association for Identification, FBI, ATF, Organization of Scientific Area Committees, and more.
Welsh also sought to cross-examine McCarthy using a report from the National Academy of
Sciences (NAS), but that too was excluded on hearsay grounds because McCarthy testified that it
was not relied upon by those in her field.
In his case, Welsh introduced testimony from Rishi’s friend Tim Barbrow, who stated
that Rishi’s pill source may have been “drying up.” He stated that the new pills Rishi was selling
were extremely expensive and that he was put off by the new price tag.
Welsh then questioned another one of Rishi’s friends, Adam Masters. Masters stated that
in the week or two leading up to Rishi’s murder, Masters noticed that Rishi sustained scrapes and
bruises to his face. Masters confronted Rishi about his injuries, and Rishi claimed that he tripped
on the stairs. Masters was skeptical of Rishi’s explanation because Rishi had numerous marks
on his face. Masters did acknowledge that it was possible Rishi could have sustained his injuries
10 by tripping on the stairs, but Masters also testified that Rishi “was seeming a little more nervous
about things, like someone may have done something to him.”
Welsh also elicited testimony from Heather Norman, who claimed to see Rishi’s pit bull
outside with somebody on the morning of January 30, 2018, a day after the Commonwealth
claimed that the Manwanis were murdered. Norman testified that the person accompanying the
dog had a “broad type of silhouette” and was approximately six feet tall, much taller than Welsh.
On cross-examination, Norman could not remember exactly which day she saw the pit bull. She
testified that when she saw the murders on the news towards the end of February, she called the
sheriff’s office to report that she saw the dog outside on January 30. She testified that she drove
that route on a regular basis and that she drove it on January 29 through January 31.
Detective Bruns, who searched Welsh’s car, testified that he did not find any guns,
gunshot residue, blood, or anything else incriminating during his search. He also noted that the
car did not appear to have been detailed recently, although it was “relatively clean.”
Welsh then sought to introduce the expert testimony of William Tobin to counter
McCarthy’s testimony that the cartridge cases found at the Manwani’s “matched” Welsh’s Buck
Mark pistol. The Commonwealth objected, arguing that Welsh attempted to call Tobin “merely
to undermine the credibility of” McCarthy. Welsh disagreed and explained that Tobin would be
introduced as an expert in forensic metallurgy and material science, and that he was qualified to
address toolmark and firearm identification.
Welsh then provided a proffer of Tobin’s credentials and his expected testimony,
including the following:
• Tobin worked for the FBI for 27 years during which time he conducted toolmark analysis and oversaw other firearm and toolmark examiners;
11 • He had conducted extensive research on the forensic application of metallurgy, including bullet and lead analysis, and has “studied toolmark and firearm and science principles that are relied upon in the area to support the practice and methodology [McCarthy] used to identify” the Buck Mark pistol;
• He has published works related to the field of forensic metallurgy;
• He oversaw a firearm manufacturing plant for years;
• He would testify that he reviewed McCarthy’s worksheets, case file, and her testimony, and would opine on McCarthy’s “methods, protocols, and procedures”;
• He would explain that McCarthy’s assertion that “[m]anufacturing does not have an impact on the ability to identify” a firearm “match” is incorrect;
• He would testify that an examiner cannot find a “match” without identifying the firearm’s subclass characteristics, which can only be identified by investigating the process through which the firearm was manufactured;
• He would expound on why consecutively manufactured firearms do not leave unique markers such that an examiner can differentiate between them;
• He would describe why “metallurgy and material science [are] controlling factor[s] in the production of firearms” and talk about tribology, which McCarthy acknowledged is a factor in the manufacturing of firearms;
• He would discuss the pertinent literature and its implications for the field of firearm and toolmark identification;
• He would address McCarthy’s assertion that the field of firearm and toolmark analysis does not have an error rate; and
• He would illustrate the flaws in McCarthy’s methodology and her determination that the cartridge cases found at the Manwani’s “match” the Buck Mark pistol.
Welsh then stated that Tobin’s knowledge would help the jury understand the evidence,
his testimony was based on empirical data and studies, and he would not opine on the credibility
of McCarthy. The trial court responded that, based on Welsh’s proffer, Tobin would
12 impermissibly express an opinion on McCarthy’s credibility. The court expounded, stating that
Welsh “said [Tobin] reviewed [McCarthy’s] case file and [was] going to comment on her work.”
After a lengthy back and forth, the trial court stated that Welsh’s proffer did not satisfy Rule
2:702 for two reasons: (1) Tobin’s testimony would constitute an opinion on the credibility of
another witness, and (2) it would “attack the underlying science” of McCarthy’s opinion, “which
the [c]ourt ha[d] already ruled on pretrial.”
Welsh asked to put Tobin on the stand outside of the presence of the jury to preserve the
record, but the trial court refused. Although Welsh claimed that Tobin’s testimony was too
technical and detailed for him to submit a proffer without Tobin taking the stand, the trial court
disagreed.
In its closing argument to the jury, the Commonwealth emphasized McCarthy’s
testimony. It told the jury, “you heard [McCarthy] testify pretty extensively and for a decent
amount of time about her analyses, the multiple analyses that she conducted on the shell casings
found at the scene of the murder, on the shell casings from the firing of that weapon back at the
lab, on the bullets founds at the scene, on the lead fragments found at the scene, on the bullets
pulled from the bodies of Rishi and Mala Manwani and the bullets test fired back at the lab.”
The Commonwealth reminded the jury that McCarthy concluded that Welsh’s Buck Mark pistol
fired the casings found at the Manwani’s residence.
The Commonwealth also focused on McCarthy’s methodology, stating that “you heard
her testify about how she conducts her examinations, how she compares those shell casings, how
she utilizes the comparison microscope, what her training is. You saw her and you heard her
explanation of what a cartridge is and how a cartridge works and what centerfire is and what
rimfire is.” The Commonwealth continued, “[y]ou saw her manipulating the firearm, explaining
13 to you how she clears it, how she makes sure it’s in operating condition before she tests it, how
it’s safe to test, how extensive her examinations are, and her comparisons are before she makes
an identification.” The Commonwealth stated that McCarthy could have concluded that her tests
were inconclusive or not a “match,” and attempted to bolster her credibility by noting how she
concluded that the barrel on the Buck Mark was not the one that expelled the cartridge casings
found at the Manwani residence.
Welsh’s closing argument also noted how important McCarthy’s testimony was, going so
far as saying the Commonwealth’s case was “built on” her testimony. Welsh criticized
McCarthy’s conclusions by emphasizing that she did not examine any consecutively
manufactured firearms. Additionally, Welsh argued that McCarthy worked backwards from her
conclusion that the Buck Mark expelled the cartridge casings to hypothesize that the barrel was
switched, flipping the scientific process on its head.
During its rebuttal argument, the Commonwealth again highlighted McCarthy’s work and
its importance to the case. The Commonwealth criticized Welsh’s cross-examination-based
criticisms of McCarthy’s methodology and conclusions as nothing more than a “red herring.”
After days of deliberation, the jury found Welsh guilty of two counts of first-degree
murder and two counts of using a firearm in the commission of a felony. The jury sentenced
Welsh to two life sentences plus an additional six years in prison.
Welsh sought review in the Court of Appeals, arguing, in pertinent part, that the trial
court improperly excluded Tobin’s testimony under Rule 2:702 and that, at a minimum, Tobin
should have been allowed to testify outside of the presence of the jury in order to create a record
for appeal. In a published opinion, a panel of the Court of Appeals, ruling on what it
characterized as the best and narrowest grounds, assumed without deciding that the trial court
14 erred by excluding Tobin’s testimony, found any such error to be harmless, and affirmed the trial
court’s judgment. See Welsh v. Commonwealth, 78 Va. App. 287 (2023).
Welsh now appeals.
II. ANALYSIS
Welsh asserts that the Court of Appeals erred by not reaching the merits of his Rule 2:702
argument and by finding any potential error to be harmless. We agree with Welsh. 1
A. Tobin’s proffered testimony was admissible
“Generally, the admissibility of evidence is within the discretion of the trial court and we
will not reject the decision of the trial court unless we find an abuse of discretion.” Midkiff v.
Commonwealth, 280 Va. 216, 219 (2010). “In evaluating whether a trial court abused its
discretion, we do not substitute our judgment for that of the trial court. Rather, we consider only
whether the record fairly supports the trial court’s action.” Carter v. Commonwealth, 293 Va.
537, 543 (2017) (quotations and alterations omitted). One way in which a circuit court can abuse
its discretion is by applying erroneous legal standards. Id. at 543-44; Lawrence v.
Commonwealth, 279 Va. 490, 496 (2010). “A lower court’s interpretation of the Rules of this
1 Welsh also argues that the trial court should have permitted him to provide a more detailed proffer by having Tobin testify outside the presence of the jury. “[W]hen testimony is rejected before it is delivered, an appellate court has no basis for adjudication unless the record reflects a proper proffer.” Wyche v. Commonwealth, 218 Va. 839, 842 (1978) (quoting Whittaker v. Commonwealth, 217 Va. 966, 968-69 (1997)). Although a party may provide such a proffer “by avowal of counsel, . . . the better practice is to permit the witness to answer the question in the absence of the jury.” Id. We are cognizant of the fact that “[t]he conduct of the trial is committed to the discretion of the trial court[,]” Watkins v. Commonwealth, 229 Va. 469, 484 (1985), but also note that a circuit court’s quest for efficiency may not subordinate a party’s right to a fair trial. See Peterson v. Castano, 260 Va. 299, 303-304 (2000). Because we find that the proffer in this case is sufficient to allow us to reach the other issues raised by Welsh, we do not address Welsh’s argument that the trial court erred in refusing to allow him to make his proffer by having Tobin testify outside of the presence of the jury. See Commonwealth v. Swann, 290 Va. 194, 196 (2015) (recognizing that we strive to “decide cases on the best and narrowest grounds available”) (internal quotation marks and citations omitted).
15 Court, like its interpretation of a statute, presents a question of law that we review de novo.”
Findlay v. Commonwealth, 287 Va. 111, 114 (2014) (quoting LaCava v. Commonwealth, 283
Va. 465, 469-70 (2012)).
Rule 2:702(b) states that
[e]xpert 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.
At issue in this case is whether an expert “opines on the credibility of another witness”
when that expert criticizes the methodology utilized by, and hence, the conclusions drawn by,
another expert.
In the broadest sense of the word “credibility,” any testimony that might cause a
factfinder to disbelieve another witness could be labeled an attack on the credibility of that other
witness. If the testimony of Witness A that the light was red causes the factfinder to disbelieve
Witness B’s testimony that the light was green, it can be said that Witness A’s testimony
undermined Witness B’s credibility. Neither Rule 2:702 nor any of our prior cases addressing
limitations on a witness commenting on the credibility of another witness have embraced such a
broad definition of “credibility.” 2
2 Rule 2:102 of the Virginia Rules of Evidence provides that the “Rules state the law of evidence in Virginia[,]” were “adopted to implement established principles under the common law and not to change any established case law rendered prior to the adoption of the Rules[,]” and that our prior decisions “whether decided before or after the effective date of the Rules of Evidence, may be argued to the courts and considered in interpreting and applying the Rules of Evidence.” See also Kent Sinclair, The Law of Evidence in Virginia § 1-2(a) (8th ed. 2024); Campos v. Commonwealth, 67 Va. App. 690, 706-07 (2017). Accordingly, we rely on our prior decisions delineating the limitations on a witness commenting on the credibility of another witness in determining the meaning and scope of Rule 2:702.
16 As we have noted previously regarding a question to a medical expert,
[i]t is well settled in this Commonwealth that the credibility of witnesses and the weight to be given to their testimony are questions exclusively for the jury. The settled law of this Commonwealth simply does not permit a defendant to ask a witness to opine whether another witness is ‘capable of lying.’ The finder of fact, in this instance the jury, must determine the witness’ veracity.
James v. Commonwealth, 254 Va. 95, 98 (1997) (citations omitted). See also Pritchett v.
Commonwealth, 263 Va. 182, 187 (2002) (“An expert witness may not express an opinion as to
the veracity of a witness because such testimony improperly invades the province of the jury to
determine the reliability of a witness.”). In effect, Rule 2:702’s prohibition is on an ad hominem
attack. Outside of the circumstances outlined in Rule 2:608,3 a witness, expert or otherwise, may
not opine on the truthfulness of another witness’ testimony by reference to the personal
characteristics of the witness. Nothing in our prior cases or in Rule 2:702(b) prohibits an expert
from testifying to facts, conflicting opinions, criticisms of methodology, or other non-character-
based circumstances that might cause a factfinder to disbelieve another witness. See Fitzgerald
v. Commonwealth, 223 Va. 615, 629-30 (1982) (recognizing a distinction between an expert
impermissibly opining on the veracity of a witness directly and providing testimony not directed
to the personal characteristics of the witness that could cause a factfinder to disbelieve that
witness).
When the witnesses involved are both experts, such a reading of the cases and of Rule
2:702 allows for the proverbial “battle of the experts” with the factfinder ultimately responsible
for determining which of the conflicting expert opinions to believe. See, e.g., Grattan v.
3 Not relevant here, Rule 2:608 provides the limited circumstances in which a witness may “attack[]” the credibility of another witness by reference to that witness’ reputation for truthfulness.
17 Commonwealth, 278 Va. 602, 617-18 (2009); Riner v. Commonwealth, 268 Va. 296, 329-30
(2004); and Mercer v. Commonwealth, 259 Va. 235, 242 (2000).
Given that the proffer of Tobin’s testimony demonstrated that he sought to criticize
McCarthy’s methodology and not to attack her personal veracity, our opinion in Grattan is
particularly instructive. Grattan represented the classic “battle of the experts” with each expert
“criticiz[ing] evaluation methods [used by] the other.” 278 Va. at 617. The fact that each expert
criticized the other’s methodology for the purpose of causing the factfinder to disbelieve that
expert’s opinion did not run afoul of the rule prohibiting an expert from opining on the
credibility of another witness or otherwise render the opinions inadmissible. See Kent Sinclair,
The Law of Evidence in Virginia § 13-16 (8th ed. 2024) (“A challenge to an expert’s methods
and determinations, even when made by other experts, does not render inadmissible expert
opinion based upon those methods . . . .”) (citations and footnotes omitted). Rather, the warring
opinions created a question of fact to be resolved by the factfinder. Grattan, 278 Va. at 617-18;
Sinclair, supra, at § 13-16 (“Disagreement among expert witnesses does not nullify the probative
value of their testimony. It is for the jury to determine what weight to accord the testimony of
each expert.”) (citations and footnotes omitted). As a result, the trial court erred in concluding
that Tobin’s proffered testimony was prohibited by Rule 2:702. 4
4 The well-established principle that conflicting expert opinions raise a question of fact for the trier of fact demonstrates the error in the trial court’s alternative ground, that Tobin’s testimony represented an attack on the science underlying McCarthy’s opinion about “which the [c]ourt ha[d] already ruled on pretrial.” In its role as evidentiary gatekeeper, the trial court permissibly determined pretrial that there was a sufficient scientific basis for McCarthy’s opinions to allow the jury to consider them. A pretrial finding that a proffered expert opinion is not “junk science” and may be considered by the jury does not turn that opinion into holy writ, immune from criticism by other experts. In refusing to allow the jury to even consider the conflicting opinions of a qualified expert in the relevant field, the trial court impermissibly invaded the province of the factfinder. As the aforementioned Professor Sinclair observed regarding another case, “by rejecting parts of the evidence from the [defense] expert which
18 Accordingly, the trial court erred by ruling that Tobin was precluded from criticizing
McCarthy’s methodology and conclusions. Having determined that the trial court’s decision to
prohibit Welsh from calling his firearms expert was error, we turn to the conclusion of the Court
of Appeals that any such error was harmless.
B. Harmless Error
Once a Virginia appellate court determines that a trial court has erred, it is statutorily
required to consider whether the error merits reversal of the trial court’s judgment, i.e., whether
the error was harmless. In pertinent part, Code § 8.01-678 provides that “[w]hen 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 . . .
[f]or any . . . defect, imperfection, or omission in the record, or for any error committed on the
trial.” Thus, we consistently have held that “Code § 8.01-678 makes ‘harmless-error review
required in all cases.’” Commonwealth v. Swann, 290 Va. 194, 200 (2015) (quoting Ferguson v.
Commonwealth, 240 Va. ix, ix (1990)).
This statutory command is no mere technicality, but rather, represents “a limitation on
the powers of this [C]ourt to reverse the judgment of the trial court—a limitation which we must
consider on every application for an appeal and on the hearing of every case submitted to our
judgment.” Commonwealth v. White, 293 Va. 411, 419-420 (2017) (quoting Walker v.
Commonwealth, 144 Va. 648, 652 (1926)). Accordingly, it is not enough for an appellant to
demonstrate that a trial court erred; to be entitled to relief, he must demonstrate that the error was
significant enough to merit reversal.
conflicted with evidence of the [Commonwealth’s] expert[], the trial court violated the principle that prohibits the judge in a jury trial from assessing the weight and credibility of the evidence.” Sinclair, supra, at § 13-16.
19 1. The harmless error standard
When a Virginia appellate court conducts harmless error review, the standard it applies
depends on the type of error being reviewed. If the error involves a constitutional issue, “the
harmless-error standard . . . ask[s] ‘whether there is a reasonable possibility that the evidence
complained of might have contributed to the conviction[.]’” Id. at 420-21 (quoting Chapman v.
California, 386 U.S. 18, 23 (1967)). Stated differently, to find a constitutional error harmless, an
appellate court must conclude, beyond a reasonable doubt, that absent the error the outcome
would have been the same. Id. at 420-21.
When, as in this case, the trial court’s error is not of a constitutional dimension, the
standard is different. An appellate court can conclude that a non-constitutional error is harmless
“if it can conclude that the error did not influence the jury or had but slight effect.”
Commonwealth v. Kilpatrick, 301 Va. 214, 216 (2022) (internal quotation marks and alterations
omitted). “To reach this conclusion, the evidence of guilt must be so overwhelming that it
renders the error insignificant by comparison such that the error could not have affected the”
outcome. Id. at 217 (citing Haas v. Commonwealth, 299 Va. 465, 467 (2021)). In a criminal
case, “that outcome is conviction.” Kotteakos v. United States, 328 U.S. 750, 764 (1946).
Although both harmless error standards are well-known and the above-cited cases are
familiar to both the bench and bar, application of the standards to individual cases often proves
difficult. This is so for many reasons. First, it is important to recognize that we refer to the
harmless error “standards,” not the harmless error “rules.” See, e.g., Kilpatrick, 301 Va. at 217;
Haas v. Commonwealth, 299 Va. 465, 467 (2021); White, 293 Va. at 420-22; and Swann, 290
Va. at 200-01. As the Court of Appeals previously has recognized, “legal standards are not
amenable to the creation of hard and fast rules.” Saal v. Commonwealth, 72 Va. App. 413, 425
20 n.6 (2020) (internal alterations omitted) (quoting Wynnycky v. Kozel, 71 Va. App. 177, 200
(2019)). Standards involve consideration of subjective criteria about which reasonable minds
can, and do, differ. Rules favor objective criteria that allow for a more straightforward
application. One academic commentator has noted that “[t]he paradigmatic ‘rule’ falls toward
the high end of the specificity spectrum; it ascribes definitive consequences to the satisfaction of
precise and determinate criteria[,]” while “[t]he paradigmatic ‘standard,’ by contrast, leaves
many application-related details unresolved.” Michael Coenen, Rules Against Rulification, 124
Yale L.J. 644, 652 (2014) (footnotes omitted); see also Antonin Scalia, The Rule of Law as a
Law of Rules, 56 U. Chi. L. Rev. 1175 (1989). Accordingly, it is difficult if not impossible to
detail a list of objective criteria to guide harmless error review in all cases. See Shinseki v.
Sanders, 556 U.S. 396, 407 (2009) (warning against courts “determining whether an error is
harmless through the use of mandatory presumptions and rigid rules rather than case-specific
application of judgment, based upon examination of the record”) (citing Kotteakos, 328 U.S. at
760).
Furthermore, harmless error review does not follow the normal appellate sequence of a
trial court making a decision that is then reviewed by an appellate court. By definition, there is
no trial court ruling on harmless error to be reviewed; rather, questions regarding harmless error
are to be answered by an appellate court in the first instance, meaning that there is no decision of
the factfinder to which an appellate court can defer.
As a result, harmless error review effectively requires an appellate court to do what it
normally does not do—engage in “fact” finding (and hypothetical “fact” finding, at that). This is
so because harmless error review is not the traditional appellate question of whether the
evidence, absent any error, was sufficient to support a conviction. White, 293 Va. at 422. It is
21 not even whether the jury reached the correct result. As the United States Supreme Court
explained in Kotteakos, harmless error review after a jury trial does not turn on whether the
defendant was “guilt[y] in fact[,]” but rather on whether he was “guilt[y] in law, established by
the judgment of laymen. . . . [T]he question is, not were [the jurors] right in their judgment,
regardless of the error or its effect upon the verdict. It is rather what effect the error had or
reasonably may be taken to have had upon the jury’s decision.” 328 U.S. at 764. Finally, and
critically, the question is not whether the members of the appellate court would have convicted
absent the error, but rather, whether the jurors involved in the decision would have done so. See
id. (“The crucial thing is the impact of the thing done wrong on the minds of other men, not on
one’s own, in the total setting.”). Given that the appellate court does not know what motivated
the actual jurors, the inquiry becomes hypothetical—what would the actual jurors, about whom
we know little to nothing, have done in a hypothetical world where the error did not occur?
The lack of objective criteria to apply and the hypothetical nature of the “fact” finding
required by harmless error review makes the task difficult and will lead to different judges
reaching different conclusions in the same case. Such disparate results, however, simply
demonstrate the legal maxim “that, for some decisions, conscientious jurists could reach different
conclusions based on exactly the same facts—yet still remain entirely reasonable.” Minh Duy
Du v. Commonwealth, 292 Va. 555, 564 (2016) (quoting Thomas v. Commonwealth, 62 Va. App.
104, 111 (2013)). 5
For the reasons that follow, we conclude that it cannot be said with confidence “that the
error did not influence the jury, or had but slight effect[.]” Clay v. Commonwealth, 262 Va. 253,
5 As a result, nothing we say should be taken as criticism of our dissenting colleagues or the Court of Appeals for concluding that the error here was harmless. The question is a close one and reasonable minds can and, in fact, have disagreed.
22 260 (2001) (quoting Kotteakos, 328 U.S. at 764). Because, at a minimum, we harbor a “grave
doubt” that the trial court’s error did not affect the outcome, we cannot conclude that the error
was harmless, and therefore, “the conviction cannot stand.” Id. (quoting Kotteakos, 328 U.S. at
765).
2. Application in this case
In applying this framework to the instant case, we begin by focusing on what the trial
court’s error denied Welsh—an expert counterbalance to McCarthy’s testimony linking Welsh to
the murder weapon. In a post-CSI world, it is not a stretch to believe that scientific evidence is
given great weight by jurors. Cf. United States v. Green, 405 F. Supp. 2d 104, 117 (D. Mass.
2005) (recognizing as inherent in expert testimony the phenomenon that “a certain patina
attaches to the testimony, running the risk that the jury, labeling it ‘scientific,’ will give it more
credence than it deserves”). Logically, that risk is higher when, as here, the jury hears only from
one expert and erroneously is denied the opportunity to hear a qualified competing expert’s
equally scientific criticisms of the methodology underlying the opinions.
Here, there is no question that McCarthy was qualified to render her opinions and offer
her conclusion that Welsh’s gun was the murder weapon to “a practical certainty[.]” Based on
the proffer, it is equally clear that Tobin was qualified to call that conclusion and the
methodology used to reach it into question. His nearly three decades of work for the FBI, his
publications and research, and his other qualifications related to firearms and toolmark analysis
should have allowed him to offer his critiques of McCarthy’s methodology, which necessarily
called her ultimate conclusion into question. Given his qualifications, we simply cannot say that
some reasonable jurors would not have found his methodological criticisms persuasive, and thus,
23 caused a discounting if not wholesale disregarding of McCarthy’s opinions, including her
conclusion that it was a “practical certainty” that Welsh’s firearm was the murder weapon. 6
Having concluded that there was a significant chance that Tobin’s opinions might have
influenced the jury to discount or discredit McCarthy’s testimony,7 we must address how that
would have affected the jury’s ultimate conclusion. To do so, we first review the importance of
McCarthy’s testimony to the Commonwealth’s case and then how it fit in with all of the other
evidence.
There can be little question that McCarthy’s testimony was central to the
Commonwealth’s case. With no witness to the killings and no video depicting them, the
Commonwealth necessarily had to rely on circumstantial evidence and inference to tie Welsh to
the murders. Perhaps the most critical piece of evidence giving rise to such an inference was
6 In support of its finding of harmless error, the Court of Appeals noted that McCarthy was subject to cross-examination and conceded that at least some of Tobin’s proffered criticisms of her analysis had some merit. From this, the Court of Appeals concluded that having Tobin actually present the jury with his criticisms would not have provided significant additional benefit to Welsh. See Welsh, 78 Va. App. at 306. We disagree. Although cross-examination has been called the “greatest legal engine ever invented for the discovery of truth[,]” California v. Green, 399 U.S. 149, 158 (1970) (quoting 5 J. Wigmore, Evidence § 1367, p. 29 (3d ed. 1940)), it does not follow that cross-examination of an expert always is a sufficient substitute for the jury hearing a countervailing expert criticizing the methodology used by the first expert. As the United States Supreme Court recognized in its seminal case regarding the admission of expert testimony, “[v]igorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but admissible evidence.” Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 596 (1993) (emphasis added). Thus, allowing Welsh one of the methods of attacking McCarthy’s expert testimony, cross-examination, was not sufficient to overcome denying him another, presenting a countervailing expert, to which he was entitled. 7 This is not to say that the members of this Court, having heard Tobin’s opinions, would have discounted or disregarded McCarthy’s opinions and conclusions. However, as noted above, we are not charged with determining what the erroneously excluded testimony would have caused us to conclude. Instead, we must determine what effect it might have had on the jury in this case. See Kotteakos, 328 U.S. at 764.
24 McCarthy’s opinion that, to “a practical certainty[,]” Welsh’s gun was the murder weapon.
Given the lack of a countervailing expert opinion, it provided the jury with a seemingly
objective, largely unchallenged piece of evidence that tied Welsh personally to the killing as
opposed to evidence that simply placed him in the area with an opportunity to have committed
the killings.
The significance of McCarthy’s testimony to the Commonwealth’s case is underscored
by the Commonwealth’s statements and actions in the trial court regarding that testimony. After
all, it was the Commonwealth that characterized McCarthy as “definitely the Commonwealth’s
most significant witness[,]” and reiterated that she was, at a minimum, “[o]ne of the most
important witnesses in this case[.]” That significance was underscored by the Commonwealth’s
repeated references to McCarthy’s testimony in its closing argument, which included multiple
references to her effectively unchallenged methodology. The Commonwealth finished its
closing argument by spending eight pages worth of transcript arguing the significance of
McCarthy’s analysis and findings to the jury, and then returned to McCarthy’s testimony in its
rebuttal argument. Specifically, regarding McCarthy, the Commonwealth argued that she
did an exceptional job in this case. She didn’t go searching for evidence to frame him. She did every reasonable test, every necessary test, every appropriate test in order to make a determination as to which firearm fired the casings and the bullets. Attacking her, attacking her credibility, attacking her testimony, attacking her findings, attacking her conclusions is a red herring.
The Commonwealth treated McCarthy’s testimony as central to its case, and its “red herring”
argument would have been difficult if not impossible to make in the same manner if Tobin had
been permitted to testify.
The conclusion that the trial court’s error went to a central portion of the
Commonwealth’s case and undermined Welsh’s ability to combat it does not inexorably lead to
25 the conclusion that the error was not harmless. Even a substantial trial court error in excluding
evidence may be deemed harmless if the “evidence of guilt [is] so overwhelming that it renders
the error insignificant by comparison such that the error could not have affected the” outcome.
Kilpatrick, 301 Va. at 217; see also Salahuddin v. Commonwealth, 67 Va. App. 190, 212 (2017).
In its brief in this Court, the Commonwealth summarizes the other inculpatory evidence
as being
Welsh’s acknowledged presence at Rishi’s home close in time to the Manwanis’ murders, his prevarications about when he left the Manwanis’ and what he did after leaving the Manwanis, his desperate need for cash, his possession of a Buck Mark firearm and then his subsequent shifting of its location throughout his family, his directive to his father to “get rid of the soda can” when his father apparently kept barrels in “ammo cans,” and his father’s assurance that he had already “taken it apart” and “threw it out,” there was overwhelming evidence that Welsh used a firearm to murder Rishi and his mother. Welsh’s inconsistent statements to police and attempts to downplay his relationships with Michael and Caitlin provide further evidence of his guilt.
Setting aside McCarthy’s opinions and any buttressing effect they may have had on the
remaining evidence,8 the Commonwealth is correct that there is substantial evidence that Welsh
is guilty of the crimes charged.
If the question before us was whether the other evidence, unaffected by McCarthy’s
testimony, was sufficient to sustain Welsh’s convictions, the answer would be yes. But that is
not the question before us. White, 293 Va. at 422. If the question before us was whether a
majority, if not all, of the members of this Court would have concluded that Welsh was guilty of
8 The Commonwealth’s position that it was a reasonable inference that the soda can conversations were coded discussions about hiding the gun barrel is more credible if one accepts that the murder weapon belonged to the Welsh family and was passed around by its members.
26 the crimes charged absent the error, the answer very well may have been yes. But that is not the
question before us. Kotteakos, 328 U.S. at 764.
The question before us is whether we have a “grave doubt” that absent the trial court’s
error, the jury would have reached the same judgment of conviction. Id. at 765. Because a
majority of this Court, considering the specific facts of this case, harbors such a “grave doubt,”
we cannot conclude that the trial court’s error was harmless.
In reaching this result, we reiterate that, because we are applying a “standard” as opposed
to a “rule,” our conclusion necessarily is tied to the specific facts of this case and, frustratingly,
our analysis cannot be taken from this case and easily applied to others. Shinseki, 556 U.S. at
407 (recognizing that harmless error review requires “case-specific application of judgment”).
Here, it is the combination of the significance of the trial court’s error when compared to the
other evidence offered by the Commonwealth, the scientific nature of the disputed evidence, the
centrality of McCarthy’s opinions to the Commonwealth’s case, Tobin’s qualifications and
proffered testimony, the dispute in and about the pertinent literature, and the other evidence
offered by Welsh, e.g., the evidence related to the dog at least suggesting another perpetrator,
that leads us to conclude that the error here was not harmless.
III. CONCLUSION
For the foregoing reasons, we reverse the judgment of the Court of Appeals and remand
the matter to the Court of Appeals with instructions to remand the matter to the trial court for
further proceedings if the Commonwealth be so advised.
Reversed and remanded.
27 JUSTICE KELSEY, with whom JUSTICE McCULLOUGH and JUSTICE CHAFIN, join,
concurring in part and dissenting in part.
I fully concur in the majority’s analysis that the trial court erred in refusing to allow
Tobin to testify. The proffered testimony violated neither Rule 2:702 nor our prior cases
prohibiting a witness from commenting on the credibility of another witness. I disagree,
however, with the majority’s conclusion that, given all of the other evidence in the case, this
error was anything other than harmless. For the reasons stated by the Court of Appeals in this
case, I would conclude that the trial court’s error was harmless. See Kuang-Ming Welsh v.
Commonwealth, 78 Va. App. 287, 304-10 (2023).
Accordingly, I respectfully dissent from the majority’s conclusion regarding harmless
error.