United States v. Chikosi Legins ( 2022 )


Menu:
  •                                      PUBLISHED
    UNITED STATES COURT OF APPEALS
    FOR THE FOURTH CIRCUIT
    No. 20-4390
    UNITED STATES OF AMERICA,
    Plaintiff - Appellee,
    v.
    CHIKOSI LEGINS,
    Defendant - Appellant.
    Appeal from the United States District Court for the Eastern District of Virginia at
    Richmond. David J. Novak, District Judge. (3:19-cr-00104-DJN-1)
    Argued: September 24, 2021                                      Decided: May 11, 2022
    Before DIAZ and RICHARDSON, Circuit Judges, and FLOYD, Senior Circuit Judge.
    Affirmed by published opinion. Judge Richardson wrote the opinion, in which Judge Diaz
    and Senior Judge Floyd joined.
    ARGUED: Charles A. Gavin, CAWTHORN, DESKEVICH & GAVIN, P.C., Richmond,
    Virginia, for Appellant.        Christopher Chen-Hsin Wang, UNITED STATES
    DEPARTMENT OF JUSTICE, Washington, D.C., for Appellee. ON BRIEF: Gregory
    B. Friel, Deputy Assistant Attorney General, Tovah R. Calderon, Appellate Section, Civil
    Rights Division, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C., for
    Appellee.
    RICHARDSON, Circuit Judge:
    Former federal prison guard Chikosi Legins was indicted for sexually assaulting a
    prisoner twice and then lying to law enforcement about it. A jury convicted Legins only
    of making a false statement to law enforcement while acquitting him of the more
    substantial sex-crime charges. Following that verdict, the district court made two decisions
    that boosted Legins’s sentence. First, it imposed an enhanced statutory maximum that was
    neither charged nor submitted to the jury. Second, it varied upward to impose the sentence
    Legins would have faced if he had been convicted of sexually abusing the prisoner.
    On appeal, Legins challenges his false-statements conviction.         We reject that
    challenge. Sufficient evidence supported that conviction, and any arguable inconsistency
    with the jury’s acquittal on other counts does not invalidate the false-statement conviction.
    Legins next argues that the judge improperly imposed an enhanced statutory maximum
    penalty based on a judicial finding not in the indictment or found by the jury. See Apprendi
    v. New Jersey, 
    530 U.S. 466
    , 490 (2000). We agree but are constrained to find the error
    harmless. See Washington v. Recuenco, 
    548 U.S. 212
    , 219–20 (2006); Neder v. United
    States, 
    527 U.S. 1
    , 4 (1999). Finally, we conclude that the court did not impose an
    unreasonable sentence.
    I.     Background
    A.     Alleged Sexual Assaults
    In May 2018, B.L., then a prisoner at Petersburg Federal Correctional Institution,
    was hanging flyers around the prison to inform inmates of upcoming events. Legins, then
    a prison guard, escorted B.L. as he travelled from unit to unit to distribute the flyers. At
    2
    6:10 PM, a surveillance video shows the two men entering a camera-free staff corridor,
    along which was a similarly camera-free administrative office. At 6:15 PM, the video
    shows them emerging from the other side of the corridor. They offer wildly different
    accounts of the intervening 5 minutes.
    That evening, B.L. reported to prison lieutenant Steven Arrant that he had been
    raped. After medical evaluation, B.L. prepared an affidavit describing his assault during
    those 5 minutes. According to B.L., Legins forced him to perform oral sex on him, and
    then anally raped him. Legins then ejaculated in his own hand and instructed B.L. to clean
    up.
    And according to B.L., this was not the first time Legins sexually assaulted him.
    Two months earlier, Legins was similarly escorting B.L. as B.L. hung flyers. While in an
    elevator alone, Legins allegedly pushed B.L. to his knees and instructed B.L. to perform
    oral sex on him, and B.L. did so. B.L. did not report this incident until after the May
    assault. He had, however, placed the sweatshirt he wore at the time, on which Legins had
    allegedly ejaculated, into a plastic bag. 1
    After completing his affidavit describing the assault, B.L. was taken to a local
    hospital, where a rape kit test was performed. The results were mixed. No physical indicia
    of forcible rape were found. But experts for both the prosecution and defense agreed that
    physical injuries are often not found. The exam did reveal toilet paper on B.L.’s anus,
    which the defense expert claimed would be unusual if the assault occurred as B.L.
    1
    According to B.L., there were also prior incidents of sexual harassment in which
    Legins masturbated in his presence.
    3
    described. But Legins’s DNA was found on a swab of B.L.’s anus, on B.L.’s jock strap,
    and on the sweatshirt B.L. kept from the March assault.
    As all of this was going on, Legins began acting suspiciously. Before B.L. was
    taken to the hospital, Legins called the medical bay seeking medicine (which he had never
    done and was not authorized to do) and made repeated calls to both the medical bay and
    the lieutenant’s office seeking B.L.’s whereabouts (despite having no official reason to do
    so). When B.L. was leaving the medical bay, the escorting officer heard someone he
    believed to be Legins shout “You’ve got to be kidding me!” A few days later, Legins asked
    a fellow guard to write a statement saying that Legins was only alone with B.L. for a
    minute, but the guard refused because he had no personal knowledge of the events of that
    day.
    Legins tells a very different story of what happened during those 5 minutes in May.
    In June 2018, he was interviewed by the FBI and the Office of Inspector General, placed
    under oath, and informed that false statements could be prosecuted under 
    18 U.S.C. § 1001
    .
    In the interview, Legins acknowledged taking B.L. into the secretary’s office—which was
    unusual and against protocol—but denied any sexual activity. He claimed that he engaged
    in “just conversation” with B.L. while he unsuccessfully tried to log onto the computer to
    print documents. 2 Legins also claimed that he has never had any sexual contact with an
    inmate at Petersburg. When asked how he would explain the presence of his DNA were it
    2
    Petersburg’s IT officer testified at trial that based on the prison’s data logs, no one
    tried to access that computer during the time of the May incident.
    4
    to be found, Legins recalled that he had masturbated in the restroom located next to the
    secretary’s office the day before the alleged sexual assault. Following the interview,
    Legins began repeating the masturbation story to his colleagues, adding a new detail: He
    had seen B.L. and another inmate unattended in that bathroom afterward, but did not report
    it even though they were in a restricted, guards-only area.
    B.      District Court Proceedings
    Legins was charged with sexual abuse of a ward 3 for the March incident. For the
    May incident, he was also charged with sexual abuse of a ward, aggravated sexual abuse, 4
    and deprivation of civil rights. 5 In a final count, Legins was charged with making false
    statements 6 during his interview. The language of this false-statement count—the focus of
    this appeal—is as follows:
    3
    
    18 U.S.C. § 2243
    (b) (“Whoever . . . knowingly engages in a sexual act with another
    person who is— (1) in official detention; and (2) under the custodial, supervisory, or
    disciplinary authority of the person so engaging; or attempts to do so, shall be fined under
    this title, imprisoned not more than 15 years, or both.”).
    4
    
    18 U.S.C. § 2241
    (a) (“Whoever . . . knowingly causes another person to engage in
    a sexual act—(1) by using force against that other person; or (2) by threatening or placing
    that other person in fear that any person will be subjected to death, serious bodily injury,
    or kidnapping; or attempts to do so, shall be fined under this title, imprisoned for any term
    of years or life, or both.”).
    5
    
    18 U.S.C. § 242
     (“Whoever . . . willfully subjects any person . . . to the deprivation
    of any rights, privileges, or immunities secured or protected by the Constitution or laws of
    the United States . . . shall be fined under this title or imprisoned not more than one year,
    or both.”). The indictment alleged that Legins’s conduct constituted “cruel and unusual
    punishment.” J.A. 26.
    6
    
    18 U.S.C. § 1001
    (a) (“[W]hoever, in any matter within the jurisdiction of the
    executive, legislative, or judicial branch of the Government of the United States, knowingly
    and willfully . . . (2) makes any materially false, fictitious, or fraudulent statement or
    representation . . . shall be fined under this title[and] imprisoned not more than 5 years . . . .
    (Continued)
    5
    On or about June 5, 2018, in the Eastern District of Virginia, the defendant,
    CHIKOSI LEGINS, knowingly and willfully made false, fictitious, and
    fraudulent statements and representations to Special Agents of the Federal
    Bureau of Investigation (FBI) and the Department of Justice (DOJ) Office of
    the Inspector General as to material facts in relation [to] a matter within the
    jurisdiction of the FBI and DOJ, agencies of the United States. Specifically,
    the defendant, CHIKOSI LEGINS: (1) falsely denied that he engaged in a
    sexual act with any inmate at any time at Federal Correctional Institution,
    Petersburg; and (2) falsely stated that on May 10, 2018, he attempted to use
    a computer and printer while he was engaged in “just conversation” with
    inmate B.L. when they were alone in an unattended office with no
    surveillance cameras. Those statements and representations by CHIKOSI
    LEGINS were false, because as CHIKOSI LEGINS then well knew, he had
    engaged in a sexual act with an inmate at Federal Correctional Institution,
    Petersburg; and on May 10, 2018, the defendant used an unattended office
    with no surveillance cameras to engage in a sexual act with B.L.
    During his arraignment, the government stated that the maximum sentence on the
    false-statement charge was 5 years. Legins pleaded not guilty and went to trial by jury. At
    trial, the jury was presented with a verdict form that asked whether Legins was guilty “as
    charged in [the false-statement count] of the Indictment,” and asked it to note “which—or
    both—of [the two alleged false statements] supports [its] guilty verdict.” J.A. 820.
    The jury convicted Legins on the false-statement count. In finding him guilty of
    that count, the jury found that both statements presented in the Indictment supported the
    verdict—that is, both lying about “engag[ing] in a sexual act with any inmate” and lying
    about using a computer and printer while “engaged in ‘just conversation’” with B.L. in the
    office. J.A. 820. But the jury acquitted Legins on all other counts, even though they
    involved the commission of the sexual acts that the jury found Legins lied about.
    If the matter relates to an offense under chapter 109A, 109B, 110, or 117, or section 1591,
    then the term of imprisonment imposed under this section shall be not more than 8 years.”).
    6
    After denying Legins’s motion for acquittal notwithstanding the verdict, the district
    court observed that the false-statement count justified an enhanced 8-year maximum
    sentence—not the standard 5-year maximum—because Legins’s statement involved a
    sexual crime that is an “offense under Chapter 109A” of the criminal code. Legins objected
    to the increased statutory maximum sentence under Apprendi 7 and argued that the
    “misstatement” of the 5-year maximum sentence during arraignment also prejudiced him.
    The court disagreed and applied the 8-year statutory maximum. The increased statutory
    maximum also increased Legins’s guideline offense level by 4 levels to 18. See U.S.S.G.
    §2J1.2(b)(1)(A) (“If the (i) defendant was convicted under 
    18 U.S.C. § 1001
    ; and (ii)
    statutory maximum term of eight years’ imprisonment applies because the matter relates
    to . . . chapter[] 109A . . . increase by 4 levels.”). Having no criminal history, Legins then
    faced a guidelines range of 27–33 months. See U.S.S.G. Ch. 5 Pt. A. 8
    At sentencing, the court adopted the 27–33-month range and granted the
    prosecution’s requested upward variance. The court found that, while the guidelines
    accounted for the underlying May sexual abuse, 9 they did not reflect that Legins had abused
    7
    In Apprendi v. New Jersey, 
    530 U.S. 466
     (2000), the Supreme Court held that “any
    fact (other than prior conviction) that increases the maximum penalty for a crime must be
    charged in an indictment, submitted to a jury, and proven beyond a reasonable doubt.” 
    Id. at 476
     (quoting Jones v. United States, 
    526 U.S. 227
    , 243 n.6 (1999)).
    8
    If the 5-year statutory maximum had applied, the guidelines range would have
    been 15–21 months. U.S.S.G. Ch. 5 Pt. A.
    9
    Section 2J1.2(c) provides that a defendant whose lies obstruct a criminal
    investigation is to be sentenced under § 2X3.1 as an accessory after the fact if doing so
    results in a higher sentence. The proper range under § 2X3.1 varies based on the severity
    of the underlying crime. The court found that, based on Legins’s conduct in May, the range
    under § 2X3.1 would be lower than that provided in § 2J1.2, and so it did not apply.
    7
    B.L. more than once. The guidelines range did not reflect the “related conduct from the
    March offense—namely abusive sexual contact in violation of 
    18 U.S.C. § 2244
    (a)(4).”
    J.A. 1057. 10 Thus, as a “starting point” for determining the appropriate size of the variance,
    the court calculated the guideline range for this sexual offense: 15–21 months. 
    Id.
     The
    Court then added this range to that for the false-statement count, leading to a total of 42–
    54 months. The court ultimately varied upwards to 54 months “[t]o account for” the March
    offense; the extent of Legins’s “obstruction”—i.e., his calls to the medical bay and
    lieutenant’s office during the initial investigation; his “premeditation”—i.e., concocting
    alternative explanations for the presence of his DNA; and a history of similar conduct—
    i.e., prior reprimands for making lewd comments to female wards and inappropriately
    entering female-only areas. J.A. 1058.
    The court found that a 54-month sentence was appropriate given the “seriousness of
    the Defendant’s conduct” and the need to afford “adequate deterrence to Defendant and
    other correctional officers,” while still accounting for Legins’s positive role in his family,
    country, and religious community. J.A. 1058; see 
    18 U.S.C. § 3553
    (a). This appeal
    followed. We have jurisdiction under 
    28 U.S.C. § 1291
     and 
    18 U.S.C. § 3742
    (a).
    10
    While the jury acquitted Legins of sexual misconduct, the court could still find at
    sentencing that misconduct occurred. This stems from the different standards of proof
    required at those two stages. The jury must acquit the defendant unless his guilt is
    established beyond a reasonable doubt. See, e.g., In re Winship, 
    397 U.S. 358
    , 361–62
    (1970). But at sentencing, the court considers all aggravating and mitigating information
    based on the preponderance of the evidence—i.e., that it probably happened. See United
    States v. Grubbs, 
    585 F.3d 793
    , 798–99 (4th Cir. 2009).
    8
    II.    Discussion
    We consider in turn Legins’s arguments that: (1) his conviction is not supported by
    sufficient evidence; (2) his sentence was for a crime of which he was neither indicted nor
    convicted; and (3) his sentence is otherwise unreasonable.
    A.     The Evidence Supports the Conviction
    Legins first argues that his false-statement conviction was not supported by
    sufficient evidence. A § 1001 false-statement conviction requires (1) a false statement in
    a matter involving a government agency, (2) made knowingly or willfully, that is
    (3) material to the matter within the agency’s jurisdiction. United States v. Hamilton, 
    699 F.3d 356
    , 362 (4th Cir. 2012). Legins’s sufficiency challenge focuses on the first element,
    a false statement.
    The government chose to go beyond reciting the three elements in the indictment.
    See Hamling v. United States, 
    418 U.S. 87
    , 117 (1974) (stating that an indictment is
    sufficient if it (1) “contains the elements of the offense charged and fairly informs a
    defendant of the charge against which he must defend,” and (2) “enables him to plead an
    acquittal or conviction in bar of future prosecutions for the same offense”). The indictment
    includes two allegedly false statements and gives the reasons why each statement was false.
    As a result, Legins argues that the government could only convict him if they proved the
    statements were false for the reasons given in the indictment. 11 So under his theory, the
    11
    Legins argues that proving that his statements were false for a reason other than
    that alleged in the indictment would constitute an impermissible constructive amendment
    of the indictment. Cf. Stirone v. United States, 
    361 U.S. 212
    , 218–19 (1960). A
    (Continued)
    9
    government had to prove that (1) Legins’s denial of having “engaged in a sexual act with
    any inmate at any time at Federal Correctional Institution, Petersburg” was false because
    “he had engaged in a sexual act with an inmate at Federal Correctional Institution,
    Petersburg”; or (2) Legins’s statement that “on May 10, 2018, he attempted to use a
    computer and printer while he was engaged in ‘just conversation’ with inmate B.L. when
    they were alone in an unattended office with no surveillance cameras” was false because
    “on May 10, 2018, the defendant used an unattended office with no surveillance cameras
    to engage in a sexual act with B.L.” 12 Legins argues there was insufficient evidence that
    he had performed a sexual act on an inmate, so a reasonable jury could not have found that
    his statements were false for the reasons alleged in the indictment.
    “A defendant challenging the sufficiency of the evidence to support his conviction
    bears a heavy burden.” United States v. Beidler, 
    110 F.3d 1064
    , 1067 (4th Cir. 1997)
    (cleaned up). That is because it is the jury’s role to weigh conflicting evidence and find
    the truth; once that finding is made, we will not lightly disturb it. So “the relevant question
    is whether, after viewing the evidence in the light most favorable to the prosecution, any
    rational trier of fact could have found the essential elements of the crime beyond a
    constructive amendment occurs when the government or court broadens the possible bases
    for conviction beyond those presented to the grand jury. United States v. Cannady, 
    924 F.3d 94
    , 99 (4th Cir. 2019). As explained below, sufficient evidence supports finding that
    Legins committed a sexual act on an inmate and thus that his statements were indeed false
    for the reasons alleged in the indictment. We thus need not decide whether his theory—
    that the government was limited by those reasons for falsity alleged in the indictment—is
    correct as a legal matter.
    12
    Although not express, the structure of the indictment suggests that the two reasons
    for falsity are each paired with one of Legins’s two allegedly false statements.
    10
    reasonable doubt.” Jackson v. Virginia, 
    443 U.S. 307
    , 319 (1979). Viewing the evidence
    in the light most favorable to the prosecution requires drawing all reasonable inferences in
    favor of the government and assuming the jury resolved any conflicting evidence in the
    prosecution’s favor. United States v. Jeffers, 
    570 F.3d 557
    , 565 (4th Cir. 2009). Even an
    uncorroborated account of a single witness may constitute sufficient evidence. United
    States v. Clark, 
    541 F.2d 1016
    , 1017–18 (4th Cir. 1976) (per curiam); United States v.
    Shipp, 
    409 F.2d 33
    , 35–36 (4th Cir. 1969). 13
    Sufficient evidence supported finding that both of Legins’s statements were false
    for the reason charged—that is, Legins had performed a sexual act on an inmate. 14 B.L.
    testified under oath that Legins had performed a sexual act on him. Just this account, if
    found credible by the jury (and we must assume it was), may suffice. See Shipp, 
    409 F.2d at
    35–36. But the account is also significantly corroborated. An anal swab performed after
    13
    While a single witness’s uncorroborated account usually suffices, there are
    exceptions. Most notably, there is a “deeply rooted” common-law rule that a person cannot
    be convicted of perjury “solely upon the evidence of a single witness.” Weiler v. United
    States, 
    323 U.S. 606
    , 608–09 (1945). Although § 1001 bears some similarity to perjury
    (
    18 U.S.C. § 1621
    ) as they both penalize material false statements, the Supreme Court has
    made clear that the two-witness rule derives from the “special considerations” granted to
    “witnesses compelled to testify in trials at law.” Weiler, 
    323 U.S. at 609
    . Section 1001
    does not apply to witness testimony. Hubbard v. United States, 
    514 U.S. 695
    , 714–15
    (1995). Accordingly, as our sister circuits have largely held, perjury’s special two-witness
    rule does not apply to § 1001. See Fisher v. United States, 
    231 F.2d 99
    , 105–06 (9th Cir.
    1956); United States v. Killian, 
    246 F.2d 77
    , 82 (7th Cir. 1957).
    14
    The jury could convict Legins of making a false statement if either of Legins’s
    statements were false, though they ultimately concluded that both were. See United States
    v. Dennis, 
    19 F.4th 656
    , 671 (4th Cir. 2021) (“[I]ndictments charge in the conjunctive, but
    proof in the disjunctive is enough to convict”). Similarly, we may affirm if either of the
    jury’s decisions are supported by sufficient evidence. See Griffin v. United States, 
    502 U.S. 46
    , 57 (1991). We conclude that both are.
    11
    the May incident revealed Legins’s DNA, which was also found on B.L.’s clothing
    (including his jock strap). And Legins’s testimony to explain the events was contradicted
    by the prison’s IT manager, who confirmed that Legins did not try to use the computer
    during that time. Furthermore, the defendant’s request for a colleague to falsely attest to
    personal knowledge that Legins was only with B.L. for a minute also shows he was
    conscious of his wrongdoing. See Wilson v. United States, 
    162 U.S. 613
    , 621 (1896) (“The
    destruction, suppression, or fabrication of evidence undoubtedly gives rise to a
    presumption of guilt, to be dealt with by the jury.”); United States v. McDougald, 
    650 F.2d 532
    , 533 (4th Cir. 1981). So a rational juror could have concluded beyond a reasonable
    doubt that Legins performed a sexual act on B.L., making both statements false for the
    reasons alleged.
    Legins responds by pointing to a laundry list of potential alternative sources of the
    DNA found on B.L.’s anus and clothing and to alleged inconsistencies between B.L.’s
    testimony and other evidence. Neither helps his cause. While the DNA found on B.L.’s
    anus and clothing might not have come through sexual contact, it is a reasonable inference
    that it did, given the location at which some of the DNA was found (i.e., B.L.’s anus) and
    B.L.’s testimony about its source. Legins’s alleged inconsistencies fare no better. B.L.
    testified the rape “lasted like five minutes” after which B.L. cleaned himself up. This could
    not be, Legins argues, because a 5-minute rape would not have left enough time to clean
    up during the 5 minutes and 15 seconds the two were alone. But even if B.L.’s testimony
    could be read to overstate the length of the rape, it does little to undermine his testimony.
    For someone being coerced into unwanted sexual acts, 1 minute may feel like 5. The same
    12
    goes for the Legins’s arguments about the lack of physical injury and the presence of toilet
    paper. The defense’s expert only contended that she would not “anticipate” finding a lack
    of injuries and the presence of toilet paper if the event had occurred as B.L. described, not
    that B.L.’s description was impossible, and even then the jury was not required to believe
    her. We do not ask whether Legins’s position is supported by some evidence, or even most
    evidence. We may only consider whether, viewing all the evidence in the light least
    favorable to Legins, any rational juror could have found it substantial enough to convict
    him. And here, even with some contradictory evidence, there was sufficient evidence that
    Legins committed a sex act against B.L.
    Legins counters that the evidence of any sexual act during the March and May
    incidents must have been unreliable because the jury acquitted Legins on the sexual-assault
    counts.   Though framed as part of his sufficiency challenge, this is essentially an
    inconsistent-verdict argument—that the jury inconsistently concluded that Legins did not
    commit the charged sexual acts but did commit a sexual act for the false-statement count.
    We reject this contention for two reasons.
    First, even if the verdict were internally inconsistent, it would not entitle Legins to
    relief. A defendant “cannot challenge his conviction merely because it is inconsistent with
    a jury’s verdict of acquittal on another count.” United States v. Thomas, 
    900 F.2d 37
    , 40
    (4th Cir. 1990) (citing United States v. Powell, 
    469 U.S. 57
    , 83 (1984)); accord United
    States v. Godel, 
    361 F.2d 21
    , 24 (4th Cir. 1966) (upholding a § 1001 conviction despite an
    inconsistent acquittal on the underlying crime). As we have recognized, “an inconsistent
    verdict can result from mistake, compromise, or lenity, and a jury could just as likely err
    13
    in acquitting as convicting.” United States v. Louthian, 
    756 F.3d 295
    , 305 (4th Cir. 2014).
    So a reviewing court’s “assessment of the reason for the inconsistency would be based
    either on pure speculation, or would require inquiries into the jury’s deliberations that
    courts generally will not undertake.” Powell, 
    469 U.S. at 66
    . The latter is proscribed by
    the long-standing and near-absolute no-impeachment rule, which upholds the justice
    system’s “strong interest in protecting the finality of jury verdicts, encouraging open
    deliberations in the jury room, and preventing the harassment of jurors by litigants seeking
    to overturn the verdict.” Richardson v. Kornegay, 
    3 F.4th 687
    , 702 (4th Cir. 2021). That
    leaves us with nothing but “pure speculation” into the jury’s deliberations. We decline, as
    we have many times before, to engage in such speculation. See Thomas, 
    900 F.2d at 40
    ;
    Louthian, 756 F.3d at 305–06.
    Second, even if inconsistency would entitle Legins to relief, the jury’s verdict is not
    necessarily inconsistent. The sexual-abuse counts each required the jury, as an element of
    the offense, to determine whether Legins committed a “sexual act,” and the court instructed
    them on the precise legal definition of “sexual act” for each charge: either penetration or
    contact between a mouth and a penis or anus. 15 But during its discussion of the false-
    statement count, the court did not instruct the jury on the definition of “sexual act.” Nor
    did it need to: “Sexual act” was used in that count descriptively, as part of the indictment’s
    15
    For the counts of deprivation of rights under color of law, 
    18 U.S.C. § 242
    , and
    aggravated sexual abuse, 
    18 U.S.C. § 2241
    (a), “sexual act” meant only penetrative sex or
    contact between the mouth and penis. For the two counts of sexual abuse of a ward, 
    18 U.S.C. § 2243
    (b), “sexual act” meant penetrative sex, digital penetration, contact between
    the mouth and penis, or contact between the mouth and anus. So the sexual-abuse counts
    could not be predicated solely on masturbation.
    14
    factual narrative. It was not an element of the offense, so it did not carry any particular
    legal definition.
    In diligently following the jury instructions, the jury could have correctly concluded
    that “sexual act” as used in false-statement count carried its ordinary, colloquial meaning.
    That may include acts, such as masturbation, that are excluded under the legal definitions
    of “sexual act” applicable in the sexual-abuse counts. During his closing argument, defense
    counsel proposed that the jury conclude that B.L. had voluntarily masturbated Legins—
    which would somewhat explain the presence of Legins’s DNA on B.L.’s body and
    clothing. 4 Trial Transcript at 153:7–158:10, United States v. Legins, No. 3:19-cr-00104-
    DJN-1 (E.D. Va. Feb. 11, 2020), ECF No. 138. So if the jury found—as defense counsel
    asked them to—that B.L. had masturbated Legins, they could have properly convicted
    Legins on the false-statement count, despite acquitting him on sexual-abuse counts.
    Because we find sufficient evidence to support Legins’s conviction, and because the
    jury’s verdict was not necessarily inconsistent—and would not entitle Legins to relief even
    if it were—we affirm Legins’s conviction.
    B.     The District Court Committed a Harmless Apprendi Error
    Legins next challenges the district court’s conclusion that he was subject to a
    statutorily enhanced maximum sentence for his false-statement conviction. Section 1001
    sets the statutory maximum sentence at “5 years.” But if “the matter relates to an offense
    under chapter 109A . . . then the term of imprisonment imposed under this section shall be
    not more than 8 years.” 
    18 U.S.C. § 1001
    (a). Chapter 109A includes aggravated sexual
    abuse and sexual abuse of a ward, which Legins was being investigated for when he made
    15
    the false statements. So the government could have charged and convicted Legins of the
    aggravated § 1001 offense carrying the 8-year maximum—a point that Legins himself does
    not appear to contest. See Def.’s Position on Statutory Maximum, United States v. Legins,
    No. 3:19-cr-00104-DJN-1 (E.D. Va. Mar. 12, 2020), ECF No. 144 (“Hypothetically, it
    would be disingenuous of defense counsel to argue that the false statements which are
    before the court do not relate to offenses under Chapter 109(A).”). But though he could
    have been charged, he was not. And absent a charge and conviction, Legins was not subject
    to the 8-year statutory maximum sentence. Even so, we are constrained to find the district
    court’s erroneous use of this enhanced statutory maximum harmless.
    1.      Existence of Apprendi Error
    The Fifth and Sixth Amendments guarantee that, in federal courts, “[n]o person shall
    be held to answer for a capital, or otherwise infamous crime, unless on a presentment or
    indictment of a Grand Jury,” and that guilt must be determined by “an impartial jury.” U.S.
    Const. amends. V, VI. This significant safeguard of individual liberty ensures that the
    government cannot take away that liberty without the consent of two juries. It is thus
    necessary that all elements of an offense be included in the grand jury’s indictment and
    submitted to the petit jury for its guilt determination. “[A] fact is by definition an element
    of the offense . . . if it increases the punishment above what is otherwise legally prescribed.”
    United States v. Alleyne, 
    570 U.S. 99
    , 107–08 (2013). Thus, the Constitution requires that
    any fact that increases the statutory maximum, such as the increase in § 1001 from a 5- to
    16
    an 8-year maximum, be alleged in the indictment and found by a jury, just like all other
    offense elements. Apprendi, 
    530 U.S. at
    473–74. 16
    The indictment does not allege that the matter within the FBI and DOJ’s jurisdiction
    “relate[d] to” a Chapter 109A sexual-abuse offense. And the jury’s verdict mirrors the
    indictment. Even so, the district court found the relation to a Chapter 109A sexual abuse
    offense matter was implied by other facts alleged. The indictment charged that Legins
    “[f]alsely den[ied] that he engaged in a sexual act with any inmate at any time at Federal
    Correctional Institute, Petersburg.” J.A. 28. That statement was false, according to the
    indictment and the jury’s guilty verdict, because Legins had engaged in a sexual act with
    an inmate. And since sexual acts with an inmate are prohibited by Chapter 109A, the
    district court applied the enhanced statutory maximum.
    But there is a subtle yet significant mismatch between the facts alleged and found
    by the jury and the aggravating factor in § 1001. The jury found that Legins made
    statements that were false because he committed a sexual act with an inmate. Section
    1001’s aggravating factor depends on whether the “matter” within the FBI’s jurisdiction
    when the statements were made “relate[d] to” a Chapter 109A offense. Indeed, the
    statements need not be false denials of a Chapter 109A charge nor do the statements need
    to primarily concern a Chapter 109A offense. Put another way, the aggravating factor in
    16
    It is inconsequential that whether a matter “relates to an offense under Chapter
    109A” contains an inherent legal conclusion. Because it increases the statutory maximum
    sentence, it is an element of the offense. See Alleyne, 570 U.S. at 107–08. And all
    elements, even those involving a mixed question of law and fact, must be determined by
    the jury. See United States v. Gaudin, 
    515 U.S. 506
    , 512 (1995) (holding that a jury must
    decide whether a false statement under § 1001 is “material”).
    17
    § 1001 depends not on a statement’s contents, but rather the context in which it was made.
    The jury’s finding turned on the conduct that rendered the statements false, while the
    aggravating factor depends on the nature of the investigation.
    For example, imagine a situation in which a prison guard is being investigated for
    taking bribes from a particular prisoner in violation of 
    18 U.S.C. § 201
    . In actuality, the
    guard has also performed sexual acts with that prisoner. During the interview, the guard
    insists that he has never had a relationship with the prisoner, sexual or otherwise. The
    statement is a materially 17 false representation under § 1001(a)(2), which concerns an
    undiscovered Chapter 109A offense and is false because the guard has committed a Chapter
    109A offense. But the “matter” within law enforcement’s jurisdiction at the time, the
    bribery investigation, did not relate to a Chapter 109A offense, so an 8-year sentence would
    be inappropriate.
    To be sure, these are not the facts of our case; the FBI informed Legins early in the
    interview that he was being investigated for sexual abuse. But that fact was never alleged
    in the indictment, never stipulated to by the defendant, and never found by the jury. So
    neither the indictment nor the jury verdict contained sufficient facts on which to try or
    convict Legins on an 8-year aggravated § 1001 offense.
    17
    Materiality under § 1001 is a low bar. It merely requires that the statement have
    “a natural tendency to influence, or is capable influencing, the decision-making body to
    which it was addressed.” United States v. Hamilton, 
    699 F.3d 356
    , 362 (4th Cir. 2012)
    (quoting United States v. Littleton, 
    76 F.3d 614
    , 618 (4th Cir. 1996)). So a false statement
    about sexual misconduct could be material to (i.e., capable of influencing) an investigation
    unrelated to sexual misconduct.
    18
    2.     Harmless Error
    Finding this constitutional error does not end our inquiry, however, because not all
    constitutional errors require automatic reversal; instead, “[m]ost constitutional errors can
    be harmless.” Neder v. United States, 
    527 U.S. 1
    , 8 (1999) (quoting Arizona v. Fulminante,
    
    499 U.S. 279
    , 306 (1999)). 18 An error is harmless if it does not affect the defendant’s
    “substantial rights.” Fed. R. Crim. P. 52(a). For constitutional errors, the government
    18
    The government did not expressly argue in its response brief that the Apprendi
    error was harmless, first raising it in response to questioning at oral argument. As a general
    matter, we do not consider issues raised for the first time at oral argument. See United
    States v. Cornell, 
    780 F.3d 616
    , 625 n.2 (4th Cir. 2015). We have not yet addressed “head
    on” whether the government’s failure to argue harmless error in its briefing waives the
    application of the doctrine. See United States v. Bizuela, 
    962 F.3d 784
    , 798 n.10 (4th Cir.
    2020). On one hand, “[o]ur case law is clear that ‘parties cannot waive the proper standard
    of review by failing to argue it.’” United States v. Venable, 
    943 F.3d 187
    , 192 (4th Cir.
    2019) (quoting Sierra Club v. U.S. Dep’t of the Interior, 
    899 F.3d 260
    , 286 (4th Cir. 2018)).
    On the other, we have stated in dicta that harmless error review can be waived. See United
    States v. Hall, 
    858 F.3d 254
    , 280 n.8 (4th Cir. 2017). We need not resolve this tension
    here. Waiver is a discretionary doctrine, and we retain “inherent authority to consider and
    decide pertinent matters that otherwise may be ignored as abandoned or waived.” United
    States v. Holness, 
    706 F.3d 579
    , 592 (4th Cir. 2013).
    We elect to exercise that authority here, for three reasons. First, Legins’s
    presentment of the Apprendi issue in his opening brief was far from clear. Rather than
    raise it as a distinct issue under its own heading, he included it briefly alongside the Due
    Process challenge based on comments made during arraignment. The Government then
    raised harmless error in response to the arraignment issue, but not the Apprendi issue.
    Resp. Br. of Appellee 31. Given the messiness of the briefing, we are inclined to give both
    parties some leeway; we have been generous in our construction of Legins’s arguments,
    and it seems only fair to afford the government similar grace. Second, the harmless-error
    issue does not turn on complicated or unclear facts. See Holness, 706 F.3d at 592 (noting
    that looking past waiver is more appropriate where “the facts have been sufficiently
    developed to readily permit evaluation of an alternative legal theory”). Third, Legins
    argued in his opening brief that the Apprendi issue was not harmless. Op. Br. of Appellant
    22–23. So this is not a case in which a party is blindsided by an eleventh-hour argument.
    Cf. United States v. Ashford, 
    718 F.3d 377
    , 380–81 (4th Cir. 2013). Legins recognized that
    harmless error was the applicable standard and took the opportunity to argue under it.
    19
    bears the heavy duty of proving “beyond a reasonable doubt that the error complained of
    did not contribute to the [result] obtained.” Neder, 
    527 U.S. at 15
     (quoting Chapman v.
    California, 
    386 U.S. 18
    , 24 (1967)).
    But before we can determine whether the error was harmless, we need to determine
    what exactly that error was. There are two plausible options. First, we may think of it as
    a trial error: The government failed to include the aggravating factor in its indictment and
    jury instructions—omitting an essential element of its case. Or we may think of it as a
    sentencing error: Legins was properly charged and convicted of a baseline § 1001 offense,
    but the court erroneously sentenced him as if he had been convicted of the aggravated
    § 1001 offense.
    The choice between these two options dictates how we conduct harmless-error
    analysis. For trial errors—such as the omission of an element from the indictment and jury
    instructions—we ask whether the jury would have reached the same verdict had the
    element been included. See Greer v. United States, 
    141 S. Ct. 2090
    , 2100 (2021); 
    id. at 2102
     (Sotomayor, J., concurring in part) (“[A] constitutional [trial] error is harmless only
    if there is no reasonable doubt about whether it affected the jury’s actual verdict in the
    actual trial.”); Neder, 
    527 U.S. at 18
    . In Neder, for instance, the defendant was charged
    with tax fraud—making materially false statements on his tax return. 
    527 U.S. at
    6–7. But
    the trial court’s instructions to the jury omitted the necessary requirement that the
    misrepresentation be material. 
    Id. at 6, 8
    . Without that element, Neder had not been
    convicted of any crime. Yet the Supreme Court affirmed Neder’s conviction, holding that
    the constitutional error was harmless. 
    Id.
     at 19–20. This was so, the Court held, because
    20
    no rational jury could have found that Neder’s statements were immaterial given the
    “overwhelming” and “uncontroverted” evidence of materiality. 
    Id.
     at 16–18. The Court
    focused on the trial error because without the omitted element there was no crime on which
    Neder could be sentenced. Thus, Neder made clear that omitting an element is a trial error
    when the conduct is not a crime without the missing element.
    But courts struggled with whether the same held true when the missing element was
    perceived as a sentencing factor, which affects the statutory sentencing range. In such a
    case, the defendant would still be guilty of an offense without the omitted element, just one
    carrying a lower possible sentence. So we might characterize that sort of missing-element
    error as a “sentencing” error; the court erroneously sentenced the defendant based on the
    greater crime instead of the lesser crime of which he was convicted. In that case, the error
    would be harmless if “the result at sentencing would have been the same” without the error.
    United States v. Montes-Flores, 
    736 F.3d 357
    , 370 (4th Cir. 2013). In other words, had the
    defendant been sentenced based on the crime for which he was charged and convicted,
    would he have received the same sentence?
    Our circuit adopted this sentencing-error approach in United States v. Promise, 
    255 F.3d 150
    , 160 (4th Cir. 2001) (en banc). In Promise, as here, the district court imposed a
    statutory sentence enhancement (for certain drug amounts) that was not in the indictment
    or jury verdict. 
    Id. at 153
    . Because the defendant did not object below under Apprendi
    (which had not yet been decided), we reviewed for plain error. Like harmless error, plain-
    error review requires the court to assess whether the error affected the defendant’s
    “substantial rights.” Compare Fed. R. Crim. P. 52(a), with Fed. R. Crim. P. 52(b). To
    21
    decide whether the error affected Promise’s substantial rights, we began, as we do here, by
    trying to define the nature of the error.
    Chief Judge Wilkinson advocated for the trial-error approach:         Although the
    government and court had failed to include the drug quantities in Promise’s indictment and
    verdict form, the evidence of those drug amounts was overwhelming and uncontroverted.
    Promise, 
    255 F.3d at 166
     (Wilkinson, C.J., concurring in part). So, Chief Judge Wilkinson
    concluded, the jury would have reached the same verdict even if the missing sentencing
    factor had been included in the indictment and jury instructions, and there was accordingly
    no error affecting Promise’s substantial rights. 
    Id.
    But his view did not convince the court, which concluded that “the failure to charge
    a specific threshold drug quantity in the indictment or to instruct the jury regarding
    threshold drug quantity was not the error committed by the district court . . . . What was
    not valid was the sentence imposed, which exceeded the applicable maximum for the facts
    charged and proven.” 
    Id.
     at 160 n.8 (majority opinion). So the court implicitly rejected
    Neder’s trial-focused harmless-error standard, instead looking to whether the error affected
    Promise’s sentence. And because Promise received a higher sentence than he would have
    had the Apprendi error not occurred, the court found that his substantial rights were
    affected. 
    Id. at 160
    .
    In the years that followed, our circuit applied Promise’s sentencing conception of
    Apprendi errors in harmless-error analyses. See United States v. Stokes, 
    261 F.3d 496
    , 501
    (4th Cir. 2001); United States v. Chase, 
    296 F.3d 247
    , 250 (4th Cir. 2002). We did so most
    expressly in United States v. Mackins, 
    315 F.3d 399
     (4th Cir. 2003). As in Promise, the
    22
    three defendants in Mackins were convicted of drug crimes, but neither the indictment nor
    jury form indicated the quantities of drugs involved. 
    Id.
     at 404–05. When the judge
    determined those quantities at sentencing to increase the defendants’ sentences, one of the
    defendants objected. 
    Id. at 407
    . While conceding that an Apprendi error had occurred, the
    Government argued that it was harmless because the evidence of the drug amounts was
    “‘overwhelming’ and ‘uncontroverted.’” 
    Id. at 409
    . The court rejected that argument.
    Relying on Promise and cases applying it, the court held that the proper harmless-error
    inquiry for an “Apprendi sentencing error” is whether “it . . . result[ed] in a sentence greater
    than that which would otherwise have been imposed.” 
    Id. at 409
     (quoting Chase, 
    296 F.3d at 250
    ).
    But three years later in Washington v. Recuenco, 
    548 U.S. 212
     (2006), the Supreme
    Court directed that we are to apply the same harmless-error standard from Neder to cases
    involving omitted sentencing factors.       Arturo Recuenco threatened his wife with a
    handgun, for which he was charged in state court with assault with a deadly weapon. 
    Id. at 214
    . The jury found Recuenco guilty of assault and specified that the assault had been
    committed with a “deadly weapon.” But at sentencing, rather than impose a 1-year
    mandatory sentencing enhancement for assaults involving a “deadly weapon,” the judge
    imposed a 3-year mandatory enhancement for assaults involving a “firearm.” 
    Id. at 215
    .
    The Washington Supreme Court held that this was an Apprendi error because the defendant
    was sentenced on an offense of assault with a firearm even though the firearm element was
    neither charged nor found by the jury. 
    Id. at 216
    . The state supreme court then held that
    23
    this error was “structural”—i.e., can never be “harmless”—and accordingly vacated
    Recuenco’s sentence. 
    Id.
    The United States Supreme Court reversed, holding that Apprendi errors, like most
    constitutional errors, can be harmless. 
    Id. at 220
    . That top-line holding does not, of course,
    resolve how to perform the harmless-error analysis. But the Court’s rationale makes that
    answer clear. The Court began by describing Neder, in which, to reiterate, the Court held
    that the omission of an element of the offense may be harmless if it is “uncontested and
    supported by overwhelming evidence.” Neder, 
    527 U.S. at 17
    . Recuenco then explained:
    The State and the United States urge that this case is indistinguishable from
    Neder. We agree. Our decision in Apprendi makes clear that “[a]ny possible
    distinction between an ‘element’ of a felony offense and a ‘sentencing factor’
    was unknown to the practice of criminal indictment, trial by jury, and
    judgment by court as it existed during the years surrounding our Nation's
    founding.”
    
    548 U.S. at 220
     (quoting Apprendi, 
    530 U.S. at 478
    ).
    Recuenco argued that Neder’s trial-error approach did not apply because he was
    charged and convicted of one distinct crime (assault with a deadly weapon) but sentenced
    for another (assault with a firearm); this amounted, he argued, to “a directed verdict of guilt
    on an offense . . . greater than the one for which the jury convicted him.” Id. at 221. Justice
    Ginsburg, writing in dissent, made a similar argument. Unlike in Neder, “[n]o error marred
    the case presented at trial. The prosecutor charged, and the jury found Recuenco guilty of,
    a complete and clearly delineated offense: ‘assault in the second degree, being armed with
    a deadly weapon.’” Id. at 225 (Ginsburg, J., dissenting). In Justice Ginsburg’s view, the
    error occurred post-trial at sentencing. The judge’s decision to tack on the uncharged,
    24
    unsubmitted firearms enhancement, she insisted, displaced “the jury’s entirely complete
    verdict with, in essence, a conviction on an uncharged greater offense.” Id. at 228.
    But the Court was unpersuaded. Under Apprendi, the Court noted, “elements and
    sentencing factors must be treated the same for Sixth Amendment purposes.” Id. at 220
    (majority opinion). Both Recuenco and Neder were sentenced for a crime which they were
    almost, but not exactly, charged and found guilty. For Neder, the missing element was
    materiality. Id. at 221. For Recuenco, it was use of a firearm. That Recuenco’s crime,
    without the missing element, was itself a different crime did not, in the Court’s view, make
    any difference. Id. Holding otherwise, the Court noted, would produce bizarre results. If
    Washington, for instance, only criminalized “assault in the second degree while armed with
    a firearm,” then the omission of the “with a firearm” element could be harmless error under
    the standard in Neder. Id. at 221–22. To hold that Neder applied in that case, but not in
    Recuenco’s would, in the Court’s view, defy logic. Id. at 222.
    The takeaway from Recuenco is clear: The Government’s failure to include a
    sentence-enhancing factor in the indictment and jury charge should be treated exactly like
    its failure to include any other element of an offense. And the proper way to perform
    harmless-error analysis in both cases is to ask whether proof of the missing element is
    “overwhelming” and “uncontroverted.” Neder, 
    527 U.S. at
    17–18. In the years following
    Recuenco, that is how we, along with most of our sister circuits, have analyzed preserved
    Apprendi errors. United States v. Catone, 
    769 F.3d 866
    , 874 (4th Cir. 2014); accord United
    States v. Harakaly, 
    734 F.3d 88
    , 95–97 (1st Cir. 2013); United States v. Confredo, 
    528 F.3d 143
    , 156 (2d Cir. 2008); Tarver v. Banks, 541 F. App’x 434, 438–39 (5th Cir. 2013); United
    25
    States v. Kuehne, 
    547 F.3d 667
    , 681 n.4 (6th Cir. 2008); United States v. Hollingsworth,
    
    495 F.3d 795
    , 806 (7th Cir. 2007); United States v. Guerrero-Jasso, 
    752 F.3d 1186
    , 1193
    (9th Cir. 2014); United States v. King, 
    751 F.3d 1268
    , 1279–80 (11th Cir. 2014); United
    States v. Lopesierra-Gutierrez, 
    708 F.3d 193
    , 208 (D.C. Cir. 2013); contra United States
    v. Lewis, 
    802 F.3d 449
    , 456 (3d Cir. 2015) (en banc) (treating Apprendi error as a
    sentencing error and declining to apply Neder harmless-error test).
    In Catone, the defendant was charged and convicted of making false statements
    relating to federal benefits, violating 
    18 U.S.C. § 1920
    . 769 F.3d at 869. Under that statute,
    if the “benefits falsely obtained” exceed $1,000, the defendant is guilty of a felony
    punishable by up to a 5-year prison term; otherwise, they are guilty only of a misdemeanor
    punishable by no more than 1 year of imprisonment. 
    18 U.S.C. § 1920
    . Catone argued
    that, because the jury never found the amount of money he fraudulently obtained, his
    sixteen-month felony sentence violated Apprendi; the government countered that any such
    error was harmless. Catone, 769 F.3d at 872–73. We found Neder’s trial-error test applied:
    An “Apprendi error is harmless ‘where a reviewing court concludes beyond a reasonable
    doubt that the omitted element was uncontested and supported by overwhelming evidence,
    such that the jury verdict would have been the same absent the error.’” Id. at 874 (quoting
    Neder, 
    527 U.S. at 17
    ). 19
    19
    In Catone, we ultimately concluded that the Apprendi error was not harmless
    because there was not “overwhelming evidence” that Catone fraudulently acquired more
    than $1,000 in benefits. Id. at 875.
    26
    We therefore apply the harmless-error analysis supplied by Neder to the Apprendi
    error before us. To do so, we ask whether there is overwhelming and uncontroverted
    evidence that the “matter” within law enforcement’s jurisdiction when they questioned
    Legins “related” to a Chapter 109A offense. Chapter 109A is entitled “Sexual Abuse,” and
    includes 
    18 U.S.C. § 2243
    , Sexual Abuse of a Minor or Ward, one of the crimes for which
    Legins was later charged. That section prohibits any sexual act with a person who is in
    official detention under the defendant’s custodial authority. 
    18 U.S.C. § 2243
    (b). Early in
    the June 5 interview, the investigators told Legins the following: “Mr. Legins, we are here
    today to interview you as a subject to answer questions in an official joint OIG-FBI
    investigation regarding sexual abuse of an inmate at FCI Petersburg.” Digitally Recorded
    Sworn Statement of Chikosi Legins 3–4, United States v. Legins, No. 3:19-cr-00104-DJN-
    1 (E.D. Va. July 6, 2020), ECF No. 174, Attach. 1.            They then reiterated that the
    investigation “pertains to sexual abuse.” 
    Id.
     There is no dispute that the investigators made
    these statements or that they are true. 20 So the statutory sentencing enhancement is
    supported by overwhelming evidence.
    It is also uncontroverted: During his argument in the district court as to why the
    sentence enhancement should not apply, Legins conceded that “it would be disingenuous
    of defense counsel to argue that the false statements which are before the court do not relate
    to offenses under Chapter 109(A)” because the alleged false statements “were in
    20
    The transcript of the interview was not admitted into evidence, though it was used
    as an aid for the jury as they listened to the audio recording of the interview, which was
    admitted by stipulation of the parties.
    27
    connection with Counts One through Four, three of which are in Chapter 109(A).” Def.’s
    Position on Statutory Maximum at 2, United States v. Legins, No. 3:19-cr-00104-DJN-1
    (E.D. Va. Mar. 12, 2020), ECF No. 144. Because the elements of an 8-year aggravated
    § 1001 offense are supported by overwhelming and uncontroverted evidence, we conclude
    that the district court’s Apprendi error is harmless beyond a reasonable doubt.
    We acknowledge that there is something deeply unsatisfying about this result. As
    Justice Scalia observed in his partial dissent in Neder, it is bizarre that a deprivation of the
    jury right, which reflects a distrust of judges to adjudicate criminal guilt, can be set aside
    as harmless when we judges find the result sufficiently clear. 
    527 U.S. at 32
     (Scalia, J.,
    dissenting in part).    It creates an inescapable irony, “in which the remedy for a
    constitutional violation by a trial judge (making the determination of criminal guilt
    reserved to the jury) is a repetition of the same constitutional violation by the appellate
    court (making the determination of guilt reserved to the jury).” 
    Id.
     And, as Justice
    Ginsburg noted in her Recuenco dissent, it is particularly unsettling in this context, when
    the defendant, “charged with one crime . . . was convicted of another . . . sans charge, jury
    instruction, or jury verdict.” 
    548 U.S. at 229
     (Ginsburg, J., dissenting).
    Such a practice “diminishes the jury’s historic capacity ‘to prevent the punishment
    from getting too far out of line with the crime.’” 
    Id.
     (quoting United States v. Mayberry,
    
    274 F.2d 899
    , 902 (2d Cir. 1960) (Friendly, J.)). But these arguments did not prevail and
    are not the law. Instead, we are bound by the majority opinions that rejected them, opinions
    holding that (1) omission of an element is harmless if evidence of the element is
    28
    overwhelming and uncontroverted and (2) that the rule applies to all elements, including
    sentencing factors. And that resolves this case.
    We also take some solace in the fact that this rule, properly applied, is a narrow one.
    It demands the government prove, beyond a reasonable doubt, that the jury would have
    convicted—also beyond a reasonable doubt—had the aggravating factor been submitted to
    them. This requires considering both record evidence and possible exculpatory evidence
    the defendant might have submitted if the crime had been properly charged. And if there
    is any reasonable possibility that a juror might have chosen not to convict, then the error
    was not harmless. Here, the aggravating factor is conclusively shown by evidence of
    unquestioned authenticity, and Legins has conceded—even after learning of the Apprendi
    issue—that it was, as a factual matter, satisfied. So the evidence is truly “overwhelming”
    and “uncontroverted,” and Legins’s Apprendi challenge is purely formalistic and
    procedural.   We reaffirm that an Apprendi error is harmless under these narrow
    circumstances. 21
    21
    Though not entirely clear, Legins also seems to suggest that the government’s
    statement during arraignment that he was only subject to a 5-year maximum on the false-
    statement count constitutes a separate reversible error. See Op. Br. of Appellant 20. And
    Rule 11(b)(1)(H) requires that a defendant be informed of the maximum possible sentence
    before entering a guilty or nolo contedere plea, but Legins did not enter such a plea. And
    even if that rule applied here, the error would be harmless because Legins’s sentence is less
    than the 5-year maximum he was notified of during arraignment. See Bell v. United States,
    
    521 F.2d 713
    , 715 (4th Cir. 1975).
    29
    C.     The Sentence Imposed was Reasonable
    Legins also challenges his sentence as unreasonable. We review the reasonableness
    of the district court’s sentence under “a deferential abuse-of-discretion standard,” even if
    the sentence is within the guidelines range. Gall v. United States, 
    552 U.S. 38
    , 41 (2007).
    First, Legins argues that the district court erred in “departing” upwards from the
    guidelines sentence. But the district court did not grant an upward departure, it granted an
    upward variance. “Departures are enhancements of, or subtractions from, a guidelines
    calculation ‘based on a specific Guidelines departure provision.’ . . . Variances, in contrast,
    are discretionary changes to a guidelines sentencing range based on a judge's review of all
    the § 3553(a) factors . . . .” United States v. Brown, 
    578 F.3d 221
    , 225–26 (3d Cir. 2009).
    Because the district court granted a variance and not a departure, Legins’s arguments that
    U.S.S.G. §§ 2J1.2 and 5K2.7 did not justify a departure are irrelevant.
    Next, Legins argues that the district court erred in finding that he had committed
    certain sexual acts with B.L. in March and May 2018. The district court found that during
    the March incident, “Defendant at least engaged in consensual sexual contact with B.L.,
    including masturbation, that constitutes a violation of 
    18 U.S.C. § 2244
    (a)(4).”
    Memorandum Order at 13, United States v. Legins, No. 3:19-cr-00104-DJN-1 (E.D. Va.
    June 30, 2020), ECF No. 171. During the May incident, the court found that “Defendant
    consensually penetrated B.L.’s anus with his penis.” 
    Id. at 17
    .
    In selecting a sentence, a court may rely on aggravating—or mitigating—facts
    found by only a preponderance of the evidence. United States v. Grubbs, 
    585 F.3d 793
    ,
    798–99 (4th Cir. 2009). We review those factual findings for clear error. United States v.
    30
    Savage, 
    885 F.3d 212
    , 225 (4th Cir. 2018). Clear error exists when “after reviewing all the
    evidence, we are ‘left with the definite and firm conviction that a mistake has been
    committed.’” United States v. Steffen, 
    741 F.3d 411
    , 415 (4th Cir. 2013) (quoting United
    States v. May, 
    359 F.3d 683
    , 688 (4th Cir. 2004)). “When reviewing factual findings for
    clear error, ‘[w]e particularly defer to a district court’s credibility determinations, for it is
    the role of the district court to observe witnesses and weigh their credibility . . . .’” United
    States v. Palmer, 
    820 F.3d 640
    , 653 (4th Cir. 2016) (quoting United States v. Abu Ali, 
    528 F.3d 210
    , 232 (4th Cir. 2008)).
    First, there was substantial evidence that some form of sexual contact occurred
    between Legins and B.L. during the March incident. To begin with, B.L.’s testimony that
    he was forced to perform oral sex on Legins may be adequate under a preponderance
    standard even if not corroborated by physical evidence. The district court believed that
    B.L. was a credible witness, a credibility determination to which we owe great deference.
    Furthermore, surveillance footage from that day shows B.L. and Legins entering the
    unmonitored area in which the elevator was located, where they remained for 5 minutes.
    Finally, the sweatshirt that B.L. was shown to be wearing in that footage tested positive for
    Legins’s DNA. The district concluded, based on this evidence, that it was more likely than
    not that some form of sexual contact occurred between B.L. and Legins on March 16, 2018.
    We are not “left with the definite and firm conviction” that this was error.
    And, as we have already discussed, there was also substantial evidence that some
    form of sexual encounter—consensual or otherwise—occurred between Legins and B.L.
    on May 10, 2018. For the same reasons that the jury possessed sufficient evidence to
    31
    conclude beyond a reasonable doubt that such an act occurred, the court did not clearly err
    in finding so by a preponderance of the evidence.
    Finally, Legins argues that “[t]he use of acquitted crimes [i.e., the March and May
    sexual incidents] to calculate an initial guideline range deprives a defendant of his Sixth
    Amendment right to a sentence wholly authorized by the jury’s verdict.” Op. Br. of
    Appellant 35. Legins’s argument is incorrect on both the facts and the law. Factually,
    Legins is incorrect that the district court used acquitted conduct to calculate his initial
    guideline range. The district court calculated Legins’s initial guidelines range as 27–33
    months, based solely on the § 1001 false-statement conviction. He then varied upward
    based on Legins’s sexual misconduct, using the Sentencing Guidelines as a guide to
    determine how large an upward variance he should grant. 22
    Legally, Legins is incorrect that the district court’s consideration of acquitted
    conduct in calculating his sentence abridges his Sixth Amendment rights. The Supreme
    Court has expressly held the opposite, as have we. See United States v. Watts, 
    519 U.S. 148
    , 157 (1997) (“We therefore hold that a jury’s verdict of acquittal does not prevent the
    sentencing court from considering conduct underlying the acquitted charge, so long as that
    conduct has been proven by a preponderance of the evidence.”); Grubbs, 
    585 F.3d at 799
    (“[A] sentencing court may consider uncharged and acquitted conduct in determining a
    22
    This is an acceptable practice. The Guidelines “reflect the collected wisdom of
    various institutions,” United States v. Goff, 
    501 F.3d 250
    , 257 (3d Cir. 2007), and “it is fair
    to assume that the Guidelines, insofar as practicable, reflect a rough approximation of
    sentences that might achieve § 3553(a)’s objectives,” Rita v. United States, 
    551 U.S. 338
    ,
    350 (2007). Judges, when varying upward based on uncharged or acquitted conduct, may
    look to that gathered wisdom to guide their discretion.
    32
    sentence . . . .”). As the Supreme Court has discussed, district courts have broad discretion
    to impose sentences fully reflecting the defendant’s “background, character, and conduct.”
    Watts, 
    519 U.S. at 151
     (quoting 
    18 U.S.C. § 3661
    ). This discretion includes the ability to
    examine the defendant’s pertinent good conduct, as well as his pertinent bad conduct—
    which of course includes related criminal activity. Just as a judge may reduce a defendant’s
    sentence upon finding that he has a “history of service to our country” and makes
    “contributions to his religious community,” J.A. 1058, so too may the judge increase the
    defendant’s sentence upon finding that he sexually abused a ward. 23
    We thus conclude that Legins’s sentence was reasonable.
    *              *             *
    The district court erred in applying the statutory enhancement without an
    authorizing jury verdict. Doing so violated Legins’s constitutional rights under Apprendi.
    But that error does not warrant reversal under Recuenco and Neder. As Legins’s remaining
    claims for relief lack merit, the district court’s judgment is
    AFFIRMED.
    23
    Legins also appears to generally challenge the district court’s weighing of the
    § 3553(a) sentencing factors given the mitigating arguments Legins made. We find no
    error.
    33