Filed:   May 9, 2000

                   UNITED STATES COURT OF APPEALS

                       FOR THE FOURTH CIRCUIT


                            Nos. 98-5(L)
                             (CR-97-2-P)



United States of America,

                                                  Plaintiff - Appellee,

          versus


Aquilia Marcivicci Barnette,

                                                 Defendant - Appellant.



                               O R D E R



     The court amends its opinion filed May 2, 2000, as follows:

     On page 2, first full paragraph of opinion, lines 8, 9, and 11

-- a section mark is inserted before each section number of the

U.S. Code.

     On page 30, first full paragraph, line 9 -- the word “damming”

is corrected to read “damning.”

                                           For the Court - By Direction



                                           /s/ Patricia S. Connor
                                                    Clerk
PUBLISHED

UNITED STATES COURT OF APPEALS

FOR THE FOURTH CIRCUIT

UNITED STATES OF AMERICA,
Plaintiff-Appellee,

v.                                                                      No. 98-5

AQUILIA MARCIVICCI BARNETTE,
Defendant-Appellant.

UNITED STATES OF AMERICA,
Plaintiff-Appellee,

v.                                                                      No. 98-11

AQUILIA MARCIVICCI BARNETTE,
Defendant-Appellant.

Appeals from the United States District Court
for the Western District of North Carolina, at Charlotte.
Robert D. Potter, Senior District Judge.
(CR-97-23-P)

Argued: September 22, 1999

Decided: May 2, 2000

Before WIDENER and NIEMEYER, Circuit Judges, and
James H. MICHAEL, Jr., Senior United States District Judge
for the Western District of Virginia, sitting by designation.

_________________________________________________________________

Affirmed in part, vacated in part, and remanded by published opinion.
Judge Widener wrote the opinion in which Judge Niemeyer and
Senior Judge Michael joined.
COUNSEL

ARGUED: James Patrick Cooney, III, KENNEDY, COVINGTON,
LOBDELL & HICKMAN, L.L.P., Charlotte, North Carolina, for
Appellant. Robert John Erickson, UNITED STATES DEPART-
MENT OF JUSTICE, Washington, D.C., for Appellee. ON BRIEF:
George V. Laughrun, II, GOODMAN, CARR, NIXON,
LAUGHRUN & LEVINE, Charlotte, North Carolina, for Appellant.
Mark Calloway, United States Attorney, Robert J. Conrad, Jr., Assis-
tant United States Attorney, Thomas G. Walker, Assistant United
States Attorney, UNITED STATES DEPARTMENT OF JUSTICE,
Washington, D.C., for Appellee.

_________________________________________________________________

OPINION

WIDENER, Circuit Judge:

Defendant, Aquilia Marcivicci Barnette, appeals his convictions
and the district court's order sentencing him to death following his
convictions. The government indicted and tried Barnette on 11 counts
stemming from the murders of Donald Lee Allen and Robin Wil-
liams, three of which are punishable by the death penalty and are per-
tinent to this appeal: use of a firearm in a carjacking that results in
death, 18 U.S.C. § 924(c), (j); commission of a carjacking that results
in death, 18 U.S.C. § 2119(3); and use of a firearm while violating the
Interstate Domestic Violence Act that results in death, 18 U.S.C.
§ 924(c), (i). After a three-week trial in January 1998, the jury found
Barnette guilty of murdering both Allen and Miss Williams, and fol-
lowing a separate sentencing trial, the jury recommended the death
sentence for those crimes, which the court imposed. Barnette presents
11 grounds for appeal before this court, alleging errors in the guilt and
sentencing phases of the trial. We address each of these grounds in
turn.

I.

Barnette and Miss Williams began dating in 1994. The two moved
in together in Roanoke, Virginia in March 1995. A little over a year

                    2
later, their relationship soured, and Miss Williams broke up with Bar-
nette in April 1996. Barnette then left the apartment they shared in
Roanoke and returned to Charlotte, North Carolina, where he lived in
his mother's house. The break-up was not amicable, however, and
Barnette continued to attempt to resume their relationship.

Miss Williams continued to live in the apartment she had shared
with Barnette, but a friend, Benjamin Greene, was staying with her
because she was afraid to remain there alone. On April 30, 1996, Miss
Williams woke Greene up, telling him "he was here," referring to Bar-
nette. Greene looked out of the window and saw Barnette smashing
the windows of Greene's car with a baseball bat. Greene attempted
to call the police, but the telephone wires had been cut. Barnette saw
Miss Williams in the apartment and began to strike at the windows
with the bat. He then threw a fire bomb through a gap he had kicked
open in the front door, setting the apartment on fire. Barnette fled the
scene after Greene fired shots at him, and Miss Williams and Greene
escaped the flames by jumping out of a rear window. Miss Williams
was hospitalized with second and third degree burns to her hands and
arms. Miss Williams identified Barnette to the Roanoke police, who
issued a warrant for his arrest and notified the Charlotte Police
Department. The Charlotte police, however, did not arrest Barnette.

On May 20, 1996, Barnette purchased a 12-gauge shotgun in Char-
lotte using his brother's, Mario Vonkeith Barnette's, Virginia driver's
license. He returned the gun the next day and exchanged it for a semi-
automatic shotgun. He sawed off the stock and barrel of the new gun
and taped a flashlight to its barrel. On June 21, 1996, Barnette took
the gun and walked from his mother's house to the nearby intersec-
tion of Billy Graham Parkway and Morris Field Road. Donald Allen
stopped his blue Honda Prelude at that intersection shortly after mid-
night. Barnette approached Allen's car with the shotgun and ordered
Allen to get out of the car. Allen complied and also threw down his
wallet after Barnette demanded it. Barnette then forced Allen to walk
at gunpoint to a drainage ditch across the road. After reaching the
ditch, Barnette shot Allen three times in the back and left his body in
the ditch. Barnette took Allen's wallet and car and drove to Bertha
Williams', Robin Williams' mother's, house in Roanoke, Virginia.
Miss Williams had been living with her mother since the firebombing
incident.

                    3
After arriving at Mrs. Williams' house, Barnette went into the
backyard and cut the home's telephone wires. He then attempted to
enter the home though the side kitchen door, but after finding that it
was locked, he fired the shotgun into the door and kicked it in. Mrs.
Williams was inside the house holding her eight-month-old grand-
daughter when Barnette entered the house. Mrs. Williams told Miss
Williams to run, and Miss Williams ran out the front door. Barnette
entered the house, confronted Mrs. Williams, and followed Miss Wil-
liams out the front door, chasing her across the street. A neighbor,
Sonji Hill, was standing in her doorway, calling the police, when Bar-
nette ran by. Barnette saw her making the call, and from 50 feet away,
he pointed the shotgun at her and told her to hang the phone up or he
would shoot her. Miss Hill hung up the phone and retreated into her
apartment, where she called the police again.

Miss Williams fell down as she was running away from Barnette,
and he caught up with her, grabbed her by the hair, and dragged her
back to her mother's house. He told Miss Williams that he planned
on killing her and himself. Mrs. Williams came out of the house as
they returned, and Miss Williams broke free from Barnette and went
with her mother toward the house. Barnette then shot Miss Williams
twice. He fired the first shot from 10 to 12 feet away, hitting Miss
Williams in the side. The second shot, fired from four to five feet
away, hit her in the back. When Barnette fired both shots, Mrs. Wil-
liams was close enough to her daughter to touch her. Miss Williams
died from these injuries.

Barnette left the scene of the murder in Allen's car, driving to
Knoxville, Tennessee where he stole new license plates for the car.
He then drove to Charlotte, North Carolina where he abandoned the
car in a shopping center parking lot on June 24, 1996. Police officers
discovered the car that night and found the shotgun Barnette used in
the murders in a nearby dumpster. Barnette turned himself in to the
police on June 25, 1996 at his mother's house. After his arrest and
Miranda warnings, Barnette took the police to the scene of Allen's
murder and showed them where to find the body. Barnette later con-
fessed to the two murders and the carjacking. No issue is made of the
admissibility of the confession.

                    4
After Barnette's arrest, the United States asserted jurisdiction over
the case and indicted him on 11 counts stemming from the murders
and firebombing.1 The government served its notice of consideration
for the death penalty on August 7, 1997, and the guilt phase of the
trial began on January 21, 1998. No witnesses at trial disputed the
facts of the crimes, and the jury found Barnette guilty on all 11 counts
of the indictment.

The trial then moved to the sentencing phase. The government
presented two statutory aggravating factors and three non-statutory
aggravating factors supporting the death sentence for Allen's murder2
and two statutory aggravating factors and three non-statutory aggra-
vating factors supporting imposition of the death sentence for Miss
Williams' murder.3 The government called several witnesses who tes-
_________________________________________________________________

1 The government charged Barnette with the following eleven crimes:
Count 1, violating the Violence Against Women Act, 18 U.S.C.
§ 2261(a)(1), (b); Count 2, using and carrying a firearm during a crime
of violence, 18 U.S.C. § 924(c)(1); Count 3, committing arson during a
felony, 18 U.S.C. § 844(h); Count 4, providing false information during
the acquisition of a firearm, 18 U.S.C. § 922(a)(6); Count 5, making a
firearm in violation of 26 U.S.C. §§ 5861(f), 5871; Count 6, being a
felon in possession of a firearm, 18 U.S.C. §§ 922(g)(1), 924(e); Count
7, committing a carjacking that results in death, 18 U.S.C. § 2119(3);
Count 8, using and carrying a firearm during a carjacking that results in
death, 18 U.S.C. § 924(c), (i); Count 9, transporting a stolen motor vehi-
cle across state lines, 18 U.S.C. § 2312; Count 10, violating the Violence
Against Women Act, 18 U.S.C. § 2261(a)(1), (b); and Count 11, using
and carrying a firearm during the commission of a violent crime that
results in death, 18 U.S.C. § 924(c)(1), (i).

2 The statutory aggravating factors consisted of (1) committing an
offense in the expectation of receiving something of pecuniary value and
(2) using substantial planning and premeditation in the offense. The non-
statutory aggravating factors included (1) the harm caused to Allen's
family as a result of the killing, (2) the risk that Barnette would likely
be a danger in the future, and (3) the existence of multiple victims in the
crime.

3 The non-statutory aggravating factors in Miss Williams' death were
the same as those for Allen. The two statutory aggravating factors were
(1) the creation of a grave risk of death to one or more persons in addi-
tion to the victim and (2) the use of substantial planning and premedita-
tion in the commission of the crime.

                    5
tified about Barnette's past violent actions. Four members of the
Allen family and three members of the Williams family also testified
about the effect the killings had on them. In consideration of mitiga-
tion of punishment, the defense presented witnesses who described
Barnette's difficult childhood, explained that he would do well in the
structured environment of prison, and indicated he had substantial
psychological problems that contributed to his violent actions. This
mitigating testimony was given by 12 of Barnette's friends and family
members and three experts who testified that Barnette would not be
a future danger in prison. Dr. Mark Cunningham presented the most
detailed testimony on future dangerousness, providing a risk assess-
ment of Barnette and concluding that there was little likelihood Bar-
nette would commit future violent acts in prison.

In rebuttal, the government called five more witnesses, including
Dr. Scott Duncan. Dr. Duncan contested Dr. Cunningham's conclu-
sion that Barnette would not be a future danger in prison based on
three factors, the Psychopathy Checklist Revised,4 research on pre-
dicting future dangerousness, and an actuarial analysis comparing
Barnette to groups of people with characteristics similar to him. Dr.
Duncan found that Barnette was likely to be violent in the future. He
testified that, in his opinion, Barnette was a psychopath. Barnette then
asked to recall his risk assessment expert, Dr. Cunningham, to rebut
Dr. Duncan's opinion that Barnette was a psychopath, and related tes-
timony based on the Psychopathy Checklist Revised, but the district
court denied that motion.

Upon conclusion of the evidence, the jury unanimously found the
existence of all the government's statutory and non-statutory aggra-
vating factors on each of Counts 7, 8, and 11 and determined that
those factors outweighed the 29 mitigating factors that at least one
juror found on each of Counts 7, 8, and 11. Thus, the jury unani-
mously recommended the court sentence Barnette to death on the
_________________________________________________________________

4 The Psychopathy Checklist Revised is a diagnostic tool created by a
certain Canadian Dr. Hass. It consists of a list of factors that a doctor
uses to evaluate a patient. Each of these factors provides a numerical
score that the doctor eventually compares to a scale. If a patient scores
a certain number on the test, the doctor may conclude that the patient is
a psychopath.

                    6
three capital counts. Prior to sentencing, Barnette made a statement
to the court, after which, pursuant to 18 U.S.C. § 3594, the district
court sentenced Barnette to death. Barnette now appeals his death
sentence, claiming 11 errors occurred in the jury selection process,
guilt phase of the trial, sentencing phase of the trial, and imposition
of the death sentence.

II.

Barnette argues that the district court should not have excluded
potential juror Rodney Bell for cause under the standard set forth in
Witherspoon v. Illinois, 

(1968), and Wainwright v. Witt,

(1985), because Bell's views on the death penalty did
not prevent or substantially impair his ability to participate in the sen-
tencing proceedings. In Witherspoon, the Court held that a death sen-
tence would be invalid if the jury that found it had been chosen by
excluding veniremen for cause who voiced general, conscientious, or
religious objections to the death 

. The
Court clarified the Witherspoon holding in Wainwright v. Witt, noting
that the juror did not have to make it "unmistakably clear that [he]
would automatically vote against imposition of capital punishment"
to justify the trial judge excluding 

. Instead, the
Court adopted the standard that the government could not challenge
the juror for cause unless his views on capital punishment "would pre-
vent or substantially impair the performance of his duties as a juror
in accordance with his oath." 

(quoting Adams
v. Texas, 

, 45 (1980)). In Witt, the Court stressed that a
juror's answers to voir dire questions could leave the record unclear
as to his opinion of the death penalty, and in these situations, the voir
dire could leave the trial judge with a definite impression that the
juror could not fairly consider the law. 

.
Therefore, the Court held that the appellate court should defer to the
trial judge's opinion because he was able to see and hear the juror in
person. 

.

We leave the determination of whether a juror is substantially
impaired under Witt in the first instance to the discretion of the trial
judge based on the voir dire inquiry. United States v. Tipton, 

, 880 (4th Cir. 1996), cert. denied, 

(1997). Our
decision in Tipton controls our review of the trial court's decision to

                     7
allow the prosecution to exclude Bell. There, we deferred to the dis-
cretion of the trial judge when certain jurors' answers in the voir dire
inquiry were ambiguous and arguably contradictory because this
inquiry turns in a large part on assessments of demeanor and credibil-
ity we cannot 

. In the case at hand, Bell's
voir dire testimony indicated that he was unclear as to his opinion on
the death penalty. While Bell did state that he would try to follow the
law to the best of his ability and would consider the death penalty, he
also indicated that he would possibly prefer one sentencing option
over the other in that "[i]f given the two choices, I would weigh
heavily on not wanting to go the death penalty unless it was very,
very, very, very well warranted." The trial judge had the opportunity
to observe Bell and assess his answers first-hand. Under these circum-
stances, we believe that a conclusion that Bell's ability to perform his
duties was substantially impaired was justified and that the court's
decision to exclude Bell was not clearly erroneous or an abuse of dis-
cretion.

III.

Barnette also claims error in the district court's decision to allow
the government to use a peremptory challenge to exclude a black
juror, Stephany Jones, claiming the government excluded Miss Jones
because of her race in violation of Batson v. Kentucky, 

(1986). Batson forbids a prosecutor from challenging jurors solely on
account of their 

. A Batson challenge consists of
three steps: (1) the defendant must make out a prima facie case that
the peremptory challenge was based on purposeful discrimination, (2)
the burden shifts to the government to produce a race neutral explana-
tion for the peremptory challenge that is particular to the parties' case
at hand, and (3) the trial court then has the duty of deciding whether
the defendant has carried his burden and proved purposeful discrimi-
nation. Purkett v. Elam, 

, 767-68 (1995). To satisfy the
second step, the prosecution is not required to offer a persuasive or
even plausible explanation for the challenge. 

.
The explanation need only be facially valid. 

.
We allow great deference to the trial judge in making the determina-
tion as to whether the proffered reason for the challenge is race neu-
tral. See, e.g., United States v. Blotcher, 

, 731 (4th Cir.
1998) (noting that court will give the district court's decision great

                     8
deference because based on determinations of credibility) (citing Bat-

n.21).

Assuming that Barnette made a prima facie case of exclusion on
the basis of race, the prosecution offered a race-neutral explanation
for the exclusion -- that Miss Jones did not favor the death penalty.
We will not disturb the trial court's acceptance of that justification in
this case because the record supports the conclusion that it was not
pretextual. The record reveals that the juror had reservations about the
death penalty, which were noted by the trial judge. Additionally,
while one of the victims in this crime was white, Donald Lee Allen,
the other was black, Miss Williams. The prosecution also produced
many black witnesses during the sentencing phase. Finally, the parties
only struck a total of three black jurors, including Miss Jones. Bar-
nette struck one of the others and concedes that the government struck
the third for legitimate reasons. Under these circumstances, we are of
opinion that the patently neutral reason for striking Miss Jones was
not pretextual.

IV.

Barnette claims the district court committed error in the guilt phase
of his trial by failing to dismiss Counts 1, 2, 3, 10, and 11 for
improper venue. The indictment describes the relevant charges as fol-
lows: Count 1 as a violation of 18 U.S.C. § 2261(a)(1), (b) for travel-
ing across state lines and firebombing Robin Williams' apartment;
Count 2 as a violation of 18 U.S.C. § 924(c)(1) for using the firebomb
during a crime of violence described in 18 U.S.C. § 2261; Count 3 as
a violation of 18 U.S.C. § 844(h)(1) for carrying explosive materials
while violating 18 U.S.C. § 2261; Count 10 as a violation of 18
U.S.C. § 2261(a)(1), (b) for traveling across state lines and commit-
ting a crime of violence, the murder of Robin Williams; and Count
11 as a violation of 18 U.S.C. § 924(c)(1), (i)(2)(1) for using a firearm
in an act of violence, the violation of 18 U.S.C. § 2261, that results
in a murder as defined in 18 U.S.C. § 1111. To determine the appro-
priate venue for trying a crime, we must identify the conduct consti-
tuting the offense and then discuss the location of the commission of
the criminal acts. United States v. Rodriguez-Moreno, 

,
279 (1999).

                    9
With respect to Counts 1 and 10, the direct violations of
§ 2261(a)(1) of the Violence Against Women Act, Barnette argues
that the Western District of Virginia is the appropriate venue because
the crimes of violence and bodily injury occurred only in Roanoke,
not during interstate travel or in Charlotte, North Carolina. The Vio-
lence Against Women Act creates an offense when "[a] person who
travels across a State line . . . with the intent to injure, harass, or
intimidate that person's spouse or intimate partner, and who, in the
course of or as a result of such travel, intentionally commits a crime
of violence and thereby causes bodily injury to such spouse or inti-
mate partner." 18 U.S.C. § 2261(a)(1) (emphasis added). The Vio-
lence Against Women Act does not contain a separate venue
provision, thus we must look to the federal venue statutes. The lan-
guage of § 2261(a)(1) designates a crime in which the defendant
crosses state lines. Cf. United States v. Bailey, 

, 766 (4th
1997), cert. denied, 

(1997). Therefore, we find that 18
U.S.C. § 3237(a) providing that "any offense . . . begun in one district
and completed in another . . . may be . . . prosecuted in any district
in which such offense was begun, continued, or completed" applies
to these counts. Barnette began the crime in North Carolina by killing
Allen and taking his car, and completed the crime by driving into Vir-
ginia and shooting Miss Williams. Thus, the offense consisted of trav-
eling and committing a violent act. The travel occurred in both North
Carolina and Virginia; therefore, venue was proper in the district
courts of North Carolina.

Both Count 2 and Count 3 depend upon an underlying violent
crime, the violations of the Violence Against Women Act, for their
provisions to apply. The Supreme Court recently decided that
"[w]here venue is appropriate for the underlying crime of violence [a
kidnapping], so too it is for the § 924(c)(1) offense." Rodriguez-

. 18 U.S.C. § 844(h)(1) is almost identical to
§ 924(c)(1), it differs only in the fact that the defendant must use fire
or explosive with the underlying crime. Thus, we find the Court's rea-
soning in Rodriguez-Moreno applies equally to determining venue for
violations of § 844(h)(1), and the government may prosecute in the
district where the underlying crime of violence occurred. As we have
determined just above, venue for violations of § 2261(a)(1) of the
Violence Against Women Act may be in a location involved in the

                     10
interstate travel used to commit the offense. Thus, venue for Counts
2 and 3 was also proper in the Western District of North Carolina.

Count 11 is the capital charge. Barnette argues that venue for this
crime should have been in the district containing Roanoke, Virginia,
where the murder took place, pursuant to the venue provision for cap-
ital murder cases in 18 U.S.C. §§ 3235-36. Section 3235 requires the
government to try capital offenses in the county where the defendant
committed the offense. The offense in this case, however, which the
statute refers to is not the murder of Miss Williams, but a violation
of 18 U.S.C. § 924(c)(1), using or carrying a firearm during or in rela-
tion to a crime of violence. This crime is a capital offense because
§ 924(j)(1) provides that the death penalty may be imposed if the
offense results in a murder as defined in 18 U.S.C. § 1111. While the
definition of murder in § 1111 applies to determine whether the crime
is a capital offense, it does not define the crime Barnette was charged
with. Again, the government prosecuted Barnette for a violation of
§ 924(c)(1), venue for which is proper in the place where the underly-
ing crime of violence occurred. Therefore, using the same logic as
with Counts 2 and 3, we are of opinion that venue was proper in the
Western District of North Carolina because the underlying crime of
violence involved traveling across state lines from Charlotte, North
Carolina to Roanoke, Virginia.

V.

Barnette argues that the evidence does not support the guilty ver-
dict on Counts 1, 2, 3, 10, and 11 because Barnette was not Robin
Williams' "intimate partner" as defined in and required by the Vio-
lence Against Women Act. 18 U.S.C. § 2266(A) defines an "intimate
partner" under § 2261(a) as a person who "has cohabited with the
abuser as a spouse." No decision of a court of appeals has come to
our attention which has construed the meaning of the phrase "as a
spouse." The district court instructed the jury on the meaning of inti-
mate partner in the words of the statute, leaving it to them to decide
whether or not Barnette and Miss Williams lived together as spouses.
As indicated by the guilty verdict on the relevant counts, the jury did
find that Barnette lived with Miss Williams as an intimate partner
within the words of the statute.

                    11
Throughout the trial, the jury heard witnesses testify about the rela-
tionship between Barnette and Miss Williams. Witnesses described
Barnette as Miss Williams' boyfriend and the couple as being happy
and in love when they started dating. The jurors knew that Miss Wil-
liams and Barnette had a dating relationship for a period of time
before they moved in together. The witnesses also established that
Barnette and Miss Williams lived together in the same apartment that
Miss Williams only began renting when the two moved in together.
Until that time, Miss Williams had lived with her mother. The rela-
tionship continued while they shared the apartment. A neighbor
observed that during this time, Barnette drove Miss Williams' car fre-
quently, using it to take her to, and pick her up from, work.

Other witnesses established that near the end of the relationship,
Miss Williams wanted Barnette out of her apartment because they
were fighting and he was too possessive, even to the point of smelling
her panties to make sure she had not cheated on him. Barnette's
brother indicated that their troubles began after Barnette said he found
a condom and a phone number in her pocket. One witness testified
that Barnette was beating Miss Williams and forcing her to have sex
while they were still living together. Additionally, police officers
responded to calls at the apartment when Barnette and Miss Williams
were fighting because she threatened to break up with him. Mrs. Wil-
liams indicated that when her daughter did not want to live with Bar-
nette any longer, she did not know how to force him to move out,
requiring her to go to her mother's house to live on several occasions.
The jury also heard testimony that after the break-up, Barnette still
loved Miss Williams and was preoccupied with her. Finally, in the
closing arguments, both the prosecution and Barnette argued to the
jury, respectively, either that Barnette was Miss Williams' intimate
partner or that the evidence did not establish that their relationship fit
within the statutory definition.

All of these intimate details about the relationship between Bar-
nette and Miss Williams were before the jury. When the word "as" is
used as a preposition, as here, a standard definition is that "as" means
"1. [i]n the role, capacity, or function of" and "2. [i]n a manner similar
to: LIKE." See Webster's II, New Riverside University Dictionary 128
(1988). And this is the definition of "as" that we adopt in these cir-
cumstances. We are of opinion that the evidence just recited makes

                     12
it clear that the intimate relationship between Barnette and Miss Wil-
liams, although not husband and wife, was "like" that of husband and
wife. Therefore, there was no reversible error in the district court's
instructing the jury in the words of the statute.

VI.5

To set properly the discussion on the admissibility of the Psychopa-
thy Checklist Revised, which follows, we must remember that the
defendant objected to its use by way of a motion in limine to exclude
evidence resulting from the use of the list.

Barnette claims as inadmissible Dr. Duncan's testimony relying on
the Psychopathy Checklist Revised because it was not evaluated for
reliability as required by Daubert v. Merrell Dow Pharmaceuticals,
Inc., 

(1993). Barnette argues that the Psychopathy
Checklist Revised is unreliable because it has not been standardized
with respect to black inmates, and Dr. Duncan improperly used race,
wealth, age, and sex to support his opinion that Barnette was a psy-
chopath. The government contends that Daubert does not apply to the
sentencing phase of a capital trial because the Supreme Court based
its decision in Daubert on its interpretation of Fed. R. Evid. 702, and
also that the Federal Rules of Evidence do not apply to the sentencing
phase of a death penalty trial. We need not address whether Daubert
applies to sentencing hearings, because, even assuming that it does,
we find the evidence meets its standard for admissibility.

In Daubert, the Supreme Court radically changed the standard for
admissibility of scientific testimony. 

(deciding to abandon test from Frye v. United States, 

,
1014 (D.C. Cir. 1923)). Instead of requiring that the scientific com-
munity generally accept scientific evidence before a court could admit
it, the Court set out a looser, two-step gatekeeping function that a trial
court must perform when evaluating the admissibility of all expert
testimony. Kumho Tire Co. v. Carmichael, 

, 147-49
(1999) (citing 

, 593-94, 597). The trial judge
must ensure that the evidence is based on scientific knowledge, or is
_________________________________________________________________

5 Parts VI-XII, inclusive, are related only to sentencing.

                    13
reliable, and ensure that the evidence will assist the trier of fact, or
is relevant. 

. The Court provided four factors
that could be used in this evaluation, but stressed its determination
that the analysis should be flexible. 

. The
Court also noted that "[v]igorous cross examination, presentation of
contrary evidence and careful instruction on the burden of proof are
traditional and appropriate means of attacking shaky but admissible
evidence." 

. The Daubert test's gatekeeping
requirement is to ensure that the expert witness in question in the
courtroom employs the same level of intellectual vigor that character-
izes the practice of an expert in the relevant field. Kumho Tire 

. When making this determination, the trial judge must
have considerable leeway in both his reliability determination and the
means he uses to conduct it. Kumho Tire 

. We
have consistently given the trial judge's decision on whether to admit
expert testimony under Daubert great deference. See, e.g., United
States v. Dorsey, 

, 814 (4th Cir. 1995), cert. denied, 

(1995). However, we must reverse if there is a clear error
in judgment on the part of the district court. Westberry v. Gislaved
Gummi AB, 

, 261 (4th Cir. 1999).

While Barnette essentially alleges that the district court made an
improper initial determination that the Psychopathy Checklist Revised
was reliable scientific evidence, he does not contest its relevancy.
When deciding whether to admit the testimony based on the Psychop-
athy Checklist Revised, the court considered Barnette's objections to
the test, that the list had not been standardized as to the black popula-
tion (Barnette is black) or as to the post middle-age population. Bar-
nette offered two articles written by his own expert, Dr. Cunningham,
to support his position, but no other evidence. On this record, the
decision to admit the evidence was not a clear error of judgment. Tak-
ing into account the discretion afforded the district court's decision,
we are of opinion that its examination of the issue was sufficient and
that the record does not support a determination that admitting the
evidence was an abuse of discretion.

Contrary to Barnette's arguments, this case is distinguishable from
United States v. Powers, 

, 1471 (4th Cir. 1995), cert.
denied, 

(1996). In Powers, the court excluded expert
testimony because the proponent of the testimony could not rebut the

                     14
evidence from the opponent that proved its unreliability. 

. Here, Barnette only proffered two articles written by his
expert, Dr. Cunningham, as evidence disputing the scientific validity
of the Psychopathy Checklist Revised. He did not, at that time, proffer
any other evidence of unreliability. In our opinion, that evidence
alone is not sufficient to show unreliability when an expert in the field
did rely on that type of testing. Absent evidence indicating more than
such a disagreement between professionals, we do not believe the dis-
trict court needed to go further to evaluate reliability.

Finally, Barnette argues that Dr. Duncan's opinion that Barnette is
a psychopath rests on impermissible factors: race, age, and poverty.
We find no merit in this argument. Dr. Duncan used 20 characteristics
in his evaluation of Barnette under the Psychopathy Checklist
Revised, in addition to the 11 characteristics he used in his future dan-
gerousness prediction, three of which were race, age, and poverty.
Barnette does not offer sufficient reasons for us to believe that Dr.
Duncan's entire opinion was based only on these three factors, and
not upon all the other factors combined. Additionally, Dr. Duncan uti-
lized several other bases for his diagnosis of psychopathy and future
dangerousness: the Diagnostic and Statistical Manual, Fourth Edition;6
his observations of Barnette's behavior; the actuarial approach; and
the research on predicting future dangerousness. Under these circum-
stances, we do not find that Dr. Duncan based his opinion on imper-
missible factors. That such testimony was admissible, however, does
not mean that it is not subject to contradiction.

VII.

Barnette next argues that the district court erred when it instructed
the jury members that if they could not agree on either a sentence of
death or life imprisonment without release, the court would sentence
Barnette "as provided by law up to life without the possibility of
_________________________________________________________________

6 The Diagnostic and Statistical Manual, Fourth Edition, is a manual
produced by the American Psychiatric Association that provides doctors
with a system for diagnosing mental disorders. In essence, the manual
lists the symptoms or elements of various disorders, and the doctor must
find a designated number of those symptoms or elements in order to
diagnose a disorder.

                    15
release." Barnette proposed a jury instruction stating that if the jury
could not agree upon a sentence, the court would "sentence the defen-
dant to life without possibility of release." Barnette's proposed
instruction is indistinguishable from the instruction refused by the dis-
trict court in United States v. Jones, 

(1999), which
refusal was affirmed by the Supreme Court.

Were that all, we would simply affirm on the basis of Jones, but
the district court did instruct the jury that in the event of disagreement
as to punishment, it would sentence Barnette "as provided by law up
to life without the possibility of release." That instruction given by the
district court is in compliance with 18 U.S.C. §§ 3593 and 3594. The
instruction proposed by Barnette, that the court, in the event of dis-
agreement, would have been required to impose a sentence of life
imprisonment without release, is simply contrary to §§ 3593 and
3594, as just mentioned.

While Jones has made it clear that the district court was not
required to give any instruction to the jury as to punishment by the
court in the event of disagreement, the district court did not have the
benefit of that decision at the time of trial, and any procedural error
made by the district court in giving the instruction that it gave is not
only harmless under ordinary procedural rules, in our opinion, it is
harmless beyond a reasonable doubt under Chapman v. California,

(1967), if considered under the Eighth Amendment and
the Due Process Clause.

VIII.

Barnette's next argument concerns one of the government's non-
statutory aggravating factors in both murders that "[t]he defendant
caused harm to the [families] of [the victims] as a result of the impact
of the killing[s] on the [families]." The Federal Death Penalty Act
allows victim impact evidence as a non-statutory aggravating factor.
It is there described as "the effect of the offense on the victim and the
victim's family." 18 U.S.C. § 3593(a) (2). The Supreme Court has
also decided that presentation of victim impact evidence does not vio-
late the Eighth Amendment. Payne v. Tennessee, 

, 827
(1991). Barnette, however, claims that the use of the victim impact

                     16
evidence in his case was unconstitutionally vague and that its presen-
tation violated his due process rights.

A majority of the Supreme Court declined to decide whether victim
impact evidence proffered under § 3593(a) (2) was unconstitutionally
vague in United States v. Jones.7 Thus, Barnette argues that the proper
standard for review of aggravating factors is for vagueness. While
Tuilaepa v. California, 

(1994), does recognize different
standards, concerning guilt or punishment, for evaluating selection
and eligibility aggravating factors,8 the Court also recognized that
"the objectives of these two inquiries can be in some tension" and that
"the proper degrees of definition of eligibility and selection factors
often is not susceptible of mathematical 

(internal citations and quotations omitted). Therefore, the Court found
that the controlling principle in analyzing aggravating factors for
vagueness is to ensure that they do not infect the death sentence with
bias or caprice. 

. Additionally, the Court
found that the "vagueness review is quite deferential" and "a factor is
not unconstitutional if it has some common sense core of meaning
. . . that criminal juries should be capable of understanding."

, 975 (quoting Jurek v. Texas, 

,
279 (1976) (White, J., concurring in judgment)) (internal quotations
omitted).

Using the deferential standard of review and examining the pro-
ceedings for bias and caprice, we find that the victim impact aggravat-
_________________________________________________________________

7 Only four justices in Jones agreed that the victim impact evidence
was not unconstitutionally vague. Jones, 

, 4514-16
(plurality opinion) (Thomas, J.; Rehnquist, C.J.; O'Connor, J.; and Ken-
nedy, J.). Justice Scalia did not provide an opinion on this question,
instead he joined in the portion of the opinion deciding that although the
factors were "loosely drafted," their use constituted harmless error.

.

8 Addressing the differences in these factors is beyond the scope of this
opinion. For an analysis of the factors, see 

-
975. We note here that the victim impact evidence may go both to the
definition of the crime and to the sentencing. See 

. But in the case at hand, none of the evidence on the effect on the
victims' families was introduced until the sentencing phase of the trial.

                    17
ing factor in this case had a common sense core of meaning. The jury
knew it was to consider the impact on the families that resulted from
the deaths of Allen and Miss Williams from the language of the factor
alone. In these circumstances, we do not find that the victim impact
aggravating factor was unconstitutionally vague as applied in
Barnette's case.

While the jury may consider victim impact aggravating factors, the
evidence used to establish it may so infect the trial that it constitutes
a due process violation. See 

. To violate due
process, an error must be of sufficient significance that it denies the
defendant the right to a fair trial. Greer v. Miller, 

, 765
(1987). No case has come to our attention in which this court or the
Supreme Court has vacated a sentence because victim impact evi-
dence violated the limits of due process.

The prosecution presented 23 witnesses during the sentencing
phase. Seven of these witnesses were members of the victims' fami-
lies testifying about the impact the crime had on their lives. These
witnesses presented stories of the victims' childhoods, family experi-
ences, and the trauma of their deaths, and poems reflecting their deep
sadness and regret over their losses. While such evidence does have
a substantial impact, we must evaluate it in the context of the entire
sentencing hearing. Of the remaining 16 witnesses, 11 witnesses,
including many of Barnette's old girlfriends, presented testimony to
the jury accusing Barnette of rape, beatings, and abuse. In mitigation,
Barnette presented a total of 23 witnesses, including ten of his own
family members. This evidence revealed that Barnette's sad and diffi-
cult childhood affected his view of the world and women.

In this context, we cannot say that the victim impact evidence
denied the defendant a fair trial. The testimony of the victims' fami-
lies was only a small portion, approximately 83 pages or 22%, of the
prosecution's total of 271 pages of its case in aggravation that
included approximately 160 pages of testimony about Barnette's vio-
lent history. Barnette had ample opportunity to counter this testimony;
his family alone testified for approximately 194 pages, or 41%, of a
total of 471 pages in his case in mitigation. Even if the victim impact
evidence went beyond the "quick glimpses of the life" of the victim
mentioned in 

, on the whole it did not contami-

                     18
nate the proceeding. Payne does not similarly limit the presentation
of evidence on the impact of the crime upon the family. See 

. Additionally, we note that in United States v.
McVeigh, the court allowed 38 witnesses to testify in support of the
victim impact non-statutory aggravating factor at the penalty phase.

, 1216, 1221-22 (10th Cir. 1998), cert. denied, 

(1999). This testimony was poignant and emotional, covering
topics such as last contacts with the victims and the "pure love and
innocence of the children killed." 

,
1221-22. The court affirmed, finding that the verdict was not based
on passion rather than reason and that the decision of the jury was
based on a reasoned moral judgment. We make those same findings
here.

IX.

Barnette alleges that the district court committed error by submit-
ting to the jury the statutory aggravating factor that Barnette created
a grave risk to others during the murder of Robin Williams. Barnette
argues this instruction was unconstitutionally vague and the evidence
presented did not support that factor. The Federal Death Penalty Act
allows the jury to consider that the "defendant, in the commission of
the offense, or in escaping . . . knowingly created a grave risk of death
to 1 or more persons in addition to the victim of the offense" as a stat-
utory aggravating factor. 18 U.S.C. § 3592(c)(5). While courts have
interpreted state statutory versions of this aggravating factor, very few
federal courts have interpreted or applied this provision of the Federal
Death Penalty Act. Interpretations of state laws are not binding upon
our construction of a federal statute. We use the standard for evaluat-
ing aggravating factors for vagueness from Tuilaepa and look to see
that the factor had a common sense core of meaning that the jury
could 

.

While a plurality of the Supreme Court has acknowledged that this
aggravating factor may be unconstitutionally vague, depending on the
facts, Gregg v. Georgia, 

, 202 (1975), when the Court
evaluated these factors, it looked to the underlying Georgia courts'
interpretations of the factors and found that they were not vague as
applied. See 

; Proffitt v. Florida, 

,
256 (1975). The only federal courts that have examined this aggravat-

                     19
ing factor did so in a pretrial context. The first evaluated
§ 3592(c)(5)'s predecessor, 21 U.S.C. § 848(n)(5), and determined
that if the government could prove that the defendant threatened
bystanders at gunpoint, the jury could consider the aggravating factor
of grave risk to others. United States v. Walker, 

,
849-50 (N.D.N.Y. 1995). In United States v. McVeigh, the court con-
sidered the current provision in the Federal Death Penalty Act and
concluded that when read in the context of the factual allegations of
the crimes, the factor was not vague. 

, 1490 (D.
Colo. 1996).9

The district court clarified the meaning of grave risk of death in the
jury instructions, defining it as "a significant and considerable possi-
bility" and as placing other persons in a "zone of danger." This defini-
tion does not differ from the plain meaning of the statute, and we are
of opinion that the instruction was not overly broad and provided a
common sense core of understanding to the factor that the jurors
could understand. As in McVeigh, the facts of this case reveal that the
government showed that Barnette created a grave risk of death both
to Sonji Hill and to Bertha Williams.

The evidence shows that Barnette aimed his shotgun at Miss Hill
when she was standing 50 feet away and threatened to shoot her. Not
only was Miss Hill in the target zone of Barnette's conduct at that
time, she was the target. Bertha Williams was standing an arm's
length away from her daughter when Barnette shot Miss Williams
from 10 to 12 feet away. Because Barnette's shot came close enough
to Bertha Williams that any small error in aim could have wounded
or killed her in addition to her daughter, Mrs. Williams was easily in
Barnette's target zone. Under these circumstances, this aggravating
factor was not vague, and the jury had sufficient evidence to conclude
that Barnette placed both of these women at a grave risk of death.
_________________________________________________________________

9 McVeigh involved the prosecution for the bombing of the Murrah
Building in Oklahoma City, Oklahoma. The bomb caused the building to
collapse, and the court considered the force of the explosion could have
created a risk to those not physically affected by the explosion. 

. The court did, however, recommend a clarifying
instruction. 

.

                    20
X.

At the end of his sentencing trial, Barnette requested that he be per-
mitted allocution to the jury before they retired to deliberate his sen-
tence. The district court denied his request, and Barnette contends that
this denial violated his statutory and Constitutional right to allocution.
The Supreme Court has not addressed the constitutionality of a defen-
dant's right to allocution. See McGautha v. California, 

,
218 n.22 (1971) (citing Hill v. United States, 

, 429
(1961)) (noting that the Court has not addressed the issue of whether
a court's refusal to allow a defendant to allocute is a Constitutional
violation), vacated sub nom. Crampton v. Ohio, 

(1972).
Barnette relies on Federal Rule of Criminal Procedure 32(c)(3)(C)
and our decision in Ashe v. North Carolina, 

(4th Cir.
1978), cert. denied, 

(1979), as the bases for his claim.
Rule 32(c)(3)(C) states that "before imposing sentence, the court must
. . . address the defendant personally and ask the defendant if the
defendant wishes to make a statement." In Ashe, we held that the
denial of a defendant's request to allocute to the state sentencing court
was a denial of due 

. However, neither Rule
32 nor Ashe answer the question of whether a defendant has a right
to address members of the jury, as opposed to the sentencing judge,
before they retire to determine the sentence.

Recently, in Green v. French, 

(4th Cir. 1998), cert.
denied, 

(1999), we had the occasion to evaluate the
application of Ashe to a defendant's request to allocute before a sen-
tencing jury, as distinguished from the sentencing court, in a North
Carolina capital murder case. Because Green was a habeas appeal, we
found that despite Ashe, the district court did not err according to
clearly established federal law when it denied the defendant the right
to allocute to the jury. 

. Because the defendant
in Ashe requested allocution before a sentencing court judge, not a
jury, we found that the lower court's determination that the defendant
did not have a right to allocute before the jury did not contradict our
holding in Ashe. 

.

A case directly on point is United States v. Hall, 

(5th
Cir. 1988), cert. denied, 

(1999). In that case, the
point on appeal was whether or not the criminal defendant had a Con-

                     21
stitutional right to make an unsworn statement of remorse before the
jury which was not subject to cross examination. The court held that
he did not, and we follow that case. 

. Neither
are we of opinion that Rule 32 has such a requirement.

XI.

Barnette alleges that the death sentence in his case violates 18
U.S.C. § 3595(c)(2)(A) because it was imposed under the influence
of passion, prejudice, or arbitrariness. We have found little guidance
on how to interpret this provision of the Federal Death Penalty Act.
Only two Fifth Circuit cases have evaluated cases under § 3595(c)
(2)(A). See United States v. Webster, 

(5th Cir. 1998),
cert. denied, 

(1999); United States v. Hall, 

(5th Cir. 1998), cert. denied, 

(1999).
From these cases and our own reading of the statute, we look to the
record to see if these factors motivated the jury's recommendation of
the death penalty, including an analysis of the aggravating factors to
see if the jury had an abundance of evidence to support imposition of
the death penalty. See 

; 

. Barnette argues that we consider the following claimed errors in
addition to the 11 grounds raised in this appeal: the court's failure to
allow Barnette to admit his confession at trial; one witness's testi-
mony that the defendant was a "cold hearted motherfucker;" the pros-
ecutor calling Dr. Cunningham a liar in his closing; the prosecution's
use of evidence that Barnette raped Alesha Chambers when he was
not convicted of that crime; the prosecution's use of witnesses testify-
ing that Barnette sexually harassed other women; the prosecutor
stressing the jury's duty during his closing; the prosecution's mention
of God's justice during the closing; and alleged contradictions in the
jury's findings that indicate they should not have issued the death
penalty.

Upon review of the record, we do not find that the jury imposed
the death penalty under improper influence. There is no indication
that the jury did not weigh the evidence, and it had sufficient evidence
to reach its particular findings on the aggravating factors, even if the
defense is correct that the above allegations were errors, which we do
not hold. We recognize that while the proceedings must be free from
passion, prejudice, and other arbitrary factors, a death penalty case

                    22
will not be emotionless. The errors alleged by the defense did not rise
to a level that overwhelmed the proceedings and created an improper
basis for the verdict.

XII.

Finally, Barnette alleges that the district court erred when it refused
to allow Dr. Mark Cunningham to testify in surrebuttal to contest Dr.
Scott Duncan's diagnosis of Barnette as a psychopath which was
presented by the government in rebuttal.

A.

First, we must determine whether it was an error for the district
court to exclude Dr. Cunningham as a witness in surrebuttal for the
defense. Surrebuttal evidence is admissible to respond to any new
matter brought up on rebuttal. United States v. King, 

,
138 (4th Cir. 1989), cert. denied, 

(1989) (citing United
States v. Burgess, 

(4th Cir. 1982)). If the evidence
would be repetitive of prior testimony, we leave the decision to admit
surrebuttal testimony to the sound discretion of the trial judge. Bur-

. In United States v. King, we held that the
district court erred when it denied the defendant the opportunity to
present surrebuttal evidence to rebut an "obviously [ ] new 

.

Before Dr. Duncan took the stand, Barnette had presented in
response to the government's case in chief three psychological or psy-
chiatric experts10 during his case in mitigation: Dr. Faye Sultan, Dr.
Seymore Halleck, and Dr. Cunningham. Dr. Sultan used the Diagnos-
tic and Statistical Manual, Fourth Edition, to diagnose Barnette with
depression and borderline personality disorder. Dr. Halleck also pro-
vided a detailed diagnosis from the Diagnostic and Statistical Manual,
_________________________________________________________________

10 The defense also called one other psychiatrist, Dr. Sally Johnson. Dr.
Johnson was the forensic psychiatrist at the prison where the government
held Barnette pending trial. She testified that Barnette had performed so
satisfactorily in prison that she felt safe to move him into the general
prison population. Dr. Johnson, however, did not testify as to Barnette's
mental health.

                    23
Fourth Edition, finding that Barnette suffered from substance abuse,
depression, bipolar disorder, intermittent explosive disorder, and bor-
derline personality disorder. Finally, Dr. Cunningham, a psychologist
and risk assessment expert, testified about the risk of Barnette com-
mitting further violent acts in prison. After explaining his methods
and the statistical bases behind his opinion, he found that Barnette's
risk for future violence in prison was very low. Dr. Cunningham was
Barnette's last witness prior to the government's case in rebuttal.

Before the government began its case in rebuttal, Barnette moved
to bar Dr. Duncan's anticipated testimony based on the Psychopathy
Checklist Revised. The trial court denied that motion. Barnette asked
to have Dr. Cunningham in the courtroom during Dr. Duncan's testi-
mony in order for him to prepare for surrebuttal. The court also
denied this motion. Dr. Duncan then testified in rebuttal for the gov-
ernment on the issue of whether Barnette would be a future danger
if placed in prison for life. Dr. Duncan testified that Barnette was a
psychopath, basing his opinion on the Diagnostic and Statistical Man-
ual, Fourth Edition, the Psychopathy Checklist Revised, and his
observations of Barnette's behavior. Despite all the psychological tes-
timony in the case of whatever type, until the government called Dr.
Duncan to the stand, there was no diagnosis of Barnette as a psycho-
path.

To repeat, until the government introduced Dr. Duncan as its wit-
ness on rebuttal in this case, no diagnosis of Barnette as a psychopath
had entered the proceeding. Indeed the word had never been men-
tioned. Certainly, if Dr. Duncan's testimony was admissible in rebut-
tal, it was admissible in the government's sentencing case in chief to
show that the defendant would not be accommodated by life impris-
onment. When the district court held that Barnette could not introduce
the only evidence he had to contradict the evidence of Dr. Duncan,
it abused whatever discretion it may have had and committed revers-
ible error, as we will demonstrate below.

While Dr. Duncan was not a psychiatrist, he was an expert in
forensic psychology. His testimony was not brief or fleeting, instead
he provided 72 pages of testimony in exquisite detail, and we set forth
the substance of that testimony as follows.

                    24
Q. Let me turn your attention back to the conclusion
you drew that the defendant was a psychopath based upon
the PCL-R?

...

Q. Before you tell us how you came to that conclusion,
will you describe for the jury what a psychopath is?

A. Certainly. In general, a psychopath is an individual
who lacks the ability to feel at the same level and have the
intensity of what feelings are as compared to nonpsycho-
pathic individuals. Typically they are very callous, manipu-
lative, calculating, individuals that will often exploit other
people. There is research to suggest that biologically, they
do not respond to what nonpsychopaths view as fear and
anxiety which are two emotions that make up what we refer
to as remorse or guilt. The pyschopath is an individual that
has little if any ability to feel remorse or guilt for behavior
they engage in.

Q. Now, did you come to the conclusion in your report
and as you testified today that the defendant is a psycho-
path?

A. That was based on the behavior that we saw during
our interviews of the defendant, also based on a review of
the material that I had cited earlier and based on how he
eventually scored on the Hare Psychopathy Checklist
Revised.

Q. Is psychopathy a formal diagnosis?

A. No, sir, it is not a diagnosis formally listed in the
DSM-4.

Q. Why is it important to establish whether or not a
defendant is a psychopath?

           25
         A. Well, psychopaths are -- criminal psychopaths are
         twice as likely to engage in future criminal behavior when
         compared to noncriminal psychopaths. Criminal psycho-
         paths are three times as likely to engage in violent future
         criminal behavior when compared to nonpsychopathic crim-
         inals. Although in any prison population only about 20 per-
         cent of that population typically are psychopaths, they are
         responsible for over 50 percent of violent crimes that are
         committed. Also, identifying someone as a psychopath helps
         institutions to be able to place them either in a maximum or
         lower security penitentiary based on what they score out.

         ...

         Q. Now, do psychopaths look any different than other
         people?

         A. No, sir. The psychopath, probably one of their best
         assets and probably one of the greatest fears of nonpsycho-
         paths is their ability to look normal. We would all like to be
         able to think that we can pick out the psychopaths from the
         nonpsychopaths in the community, but none of us are
         immune to that. . . .

         The psychopath, as I say, has the ability to look very nor-
         mal. However, if you know what you are looking for, it is
         kind of like seeing a bowel [sic] of fruit, and you say to
         yourself, gosh that bowl of fruit looks wonderful, it looks
         very good. But when you get close to the bowel [sic] of fruit
         and pick it up you realize that it's fake fruit. And the psy-
         chopath is a lot that way. And they look very, very normal,
         but when you know what to look for, you can see things in
         their behavior, not their appearance necessarily as much as
         things in their behavior, which identify them as psycho-
         paths.

After combining the psychopathy diagnosis with an actuarial analysis
of prisoners similar to Barnette and with conclusions from research
concerning predictions of future dangerousness, Dr. Duncan con-
cluded that Barnette would be a future risk in prison.

                   26
On cross examination, Barnette's attorney examined Dr. Duncan's
use of the Psychopathy Checklist Revised in his assessment of Bar-
nette. First, Barnette's attorney noted that no other experts in the trial
had used that form for diagnosis. He then questioned Dr. Duncan
extensively about an article criticizing use of the Psychopathy Check-
list Revised with black populations and mandating that doctors use it
with caution when evaluating those populations.

At the conclusion of Dr. Duncan's testimony, Barnette moved to
recall Dr. Cunningham to testify in surrebuttal to Dr. Duncan's psy-
chopathy testimony. The government objected on the grounds that the
defense should have presented their evidence concerning the psychop-
athy diagnosis and use of the Psychopathy Checklist Revised during
the testimony of their expert witnesses. Barnette countered by arguing
that he had a right to rebut the evidence Dr. Duncan presented and
that his witness would critique the Psychopathy Checklist Revised as
invalid, irresponsible science. The court denied the motion, finding
that Barnette's attorney had cross examined Dr. Duncan on the valid-
ity of the Psychopathy Checklist Revised using an article criticizing
the Psychopathy Checklist Revised and concluding that Barnette had
nothing additional to contribute. Barnette then proffered the testimony
Dr. Cunningham would have provided on surrebuttal, including: the
invalidity of the Psychopathy Checklist Revised as applied to blacks;
the inappropriateness of the use of the test on Barnette; Dr. Cunning-
ham's own application of the Psychopathy Checklist Revised to Bar-
nette, including a different score on the test that would not qualify
Barnette as a psychopath; Dr. Cunningham's opinion that Barnette
was not a psychopath; and his knowledge of no other studies address-
ing the validity of the Psychopathy Checklist Revised's application to
the black population, other than the one the defense had already intro-
duced.

We are of opinion that the testimony the defense sought to intro-
duce in surrebuttal was not cumulative or repetitive. Despite
Barnette's attorney's cross examination of Dr. Duncan on the validity
of Dr. Duncan's technique, questions from an attorney are not nearly
so effective as the testimony of a qualified expert witness. And this
must be especially true when, as here, the subject is the highly techni-
cal and specialized subject of the condition of a man's mind. Addi-
tionally, Dr. Cunningham would have testified about the fact that he

                     27
scored Barnette differently with the Psychopathy Checklist Revised
and about his own diagnostic finding that Barnette was not a psycho-
path. On this record, simple fairness required that Barnette have the
ability to rebut the new evidence the government's expert introduced
in rebuttal. See 

.

B.

While it was error for the court to refuse to allow Dr. Cunningham
to testify in surrebuttal, we cannot reverse a sentence of death "on
account of any error which can be harmless . . . where the Govern-
ment establishes beyond a reasonable doubt the error was harmless."
18 U.S.C. § 3595(c)(2); see United States v. Jones, 

, 4516 (1999) (using § 3595(2) standard for harmless error to
evaluate error caused by loosely drafted aggravating factors). Section
3595(2) incorporates the same standard for harmless error review as
that used to evaluate direct appeals of Constitutional errors. See, e.g.,
Chapman v. California, 

, 23-24 (1967). We use a different
standard, considering whether the exclusion had a substantial or inju-
rious effect on the jury's selection of the death sentence, in review of
trial-type Constitutional error in review of habeas cases. See Brecht
v. Abrahamson, 

, 638 (1993).

To evaluate the harmlessness of this error, we look to cases
addressing the importance of psychological testimony. In Satterwhite
v. Texas, the Supreme Court found that the introduction of inadmissi-
ble psychiatric testimony that the defendant would be a future danger
was not harmless beyond a reasonable doubt. 

, 258
(1988). The Court reached this determination after noting that the
doctor who testified was the only specialist in psychiatry to take the
stand and provide this type of testimony. 

.
Other witnesses had testified about the defendant's violent tendencies,
but not with the impact of a medical expert. 

. Additionally, the Court noted that the doctor stressed that the
defendant had no conscience and was a sociopath beyond the reach
of rehabilitation and that the prosecution called attention to the doc-
tor's conclusions in his closing arguments. 

-60.

In Ake v. Oklahoma, the Court found that because psychiatric evi-
dence was so important, a defendant has a Constitutional right to a

                     28
government-provided doctor if his mental state will be a significant
factor at trial. 

, 83 (1985). While the Court did not engage
in a harmless error analysis, it did note that having two views on a
mental health issue ensures both that the defendant receives a fair
adjudication of the case and that the jury does not erroneously impose
any punishment. 

. The Court also noted in such
cases its own reliance on the assumption that it had before it "both the
views of the prosecutor's psychiatrists and the`opposing views of the
defendant's doctors.'" 

. Barnette, in the case at
hand, was deprived of the "opposing views of the defendant's doc-
tors."

In United States v. MacClosky, we found that excluding key evi-
dence, which was psychiatric evidence, was not harmless error when
the witness would have offered detailed testimony vital to the
defense's ability to explain the damaging testimony. 

,
479 (4th Cir. 1982). In two other cases, we evaluated the exclusion
of psychiatric evidence under the lesser harmless error standard on
habeas review. In Boyd v. French, the defendant wanted to introduce
expert testimony that his childhood and history of personal loss made
him more likely to kill a family member. 

, 326-27 (4th
Cir. 1998), cert. denied, 

(1999). But, in the context of
a crime involving indisputable premeditation, we concluded that the
error was harmless because the court did not consider that the jury
would have been substantially influenced by the excluded testimony.

. In Tuggle v. Netherland, we found that the
erroneous exclusion of psychiatric expert testimony showing the
defendant would not be a danger in the future was harmless error
because the excluded evidence did not have a substantial or injurious
effect or influence. 

, 1393, 1395, 1396 (4th Cir. 1996),
cert. denied, 

(1996). However, we reached that determi-
nation in Tuggle while using the lesser standard for harmless error
review, and in arriving at our conclusion, we considered that the
record was unimpeachable as to the jury finding of vileness, which
also supported a sentence of death.11
_________________________________________________________________

11 In Virginia, the jury need only find one valid aggravating factor, and
so long as the invalid factor did not influence the valid factor, any error
is harmless. 

. We found that the future danger-
ousness issue did not relate to or rebut the remaining valid finding of
vileness and upheld the death sentence. 

.

                     29
After an analysis of the record and authorities, we are of opinion
that the failure to allow Dr. Cunningham to testify in surrebuttal as
to the psychopathy evidence was not harmless error and requires a
new sentencing hearing. As the Supreme Court has acknowledged,
psychiatric evidence is an important part of many trials, like this one.

. Additionally, the Court has recognized the princi-
ple that leaving this testimony unanswered can have a devastating
effect on a defendant. 

. In the case at hand, the tes-
timony of Dr. Duncan was as damning as it could be. By excluding
Dr. Cunningham's testimony, the district court left the jury with the
unrebutted expert opinion that Barnette was a psychopath who felt no
remorse or guilt, and that he resembled a fake bowl of fruit. Only by
cross examination was Barnette able to cast any doubt on this testi-
mony by questioning Dr. Duncan on the validity of his methods. In
such a case, cross examination is poor substitution for a live expert
witness. The defense was excluded from the testimony of its expert
to counter Dr. Duncan's opinions.

Again, the record shows that Dr. Duncan's analysis in rebuttal, for
the first time in the case, introduced the diagnosis of Barnette as a
psychopath. Barnette should have had the opportunity to introduce
"the opposing views of the defendant's doctor[ ]." 

.
We thus conclude that there is a reasonable possibility that the exclu-
sion of the evidence of Dr. Cunningham might have contributed to the
sentence of death and that the government has not established beyond
a reasonable doubt that the error in excluding such evidence was
harmless beyond a reasonable doubt. 18 U.S.C. § 3595(c)(2), Chap-
man v. California, 

, 24 (1966).

XIII.

In summary, the conviction of Barnette, as distinguished from his
sentence, is, in all respects, affirmed.

The sentence of death of Barnette is vacated, and the case is
remanded for a new sentencing hearing, because of the exclusion of
the testimony of Dr. Cunningham, as mentioned in this opinion.

AFFIRMED IN PART, VACATED IN PART, AND REMANDED

                    30