HENLEY, Senior Circuit Judge.
We visit this habeas case for the second time. Michael Ray Orndorff, James William Holmes, Hoyt Franklin Clines, and Darryl V. Richley (petitioners) appeal from the judgment of the district court denying their requests for habeas relief under 28 U.S.C. § 2254. Petitioners were convicted of capital murder and were sentenced to suffer the death penalty. The district court found that at trial petitioners were denied their sixth amendment right to confront the witnesses against them. The court granted petitioners partial relief, finding that this error was harmless with respect to their convictions and affected only their death sentences. Respondent Lockhart concedes the error at trial, but contends this error was harmless with respect to the convictions and sentences. We affirmed that part of the decision denying relief and remanded for further consideration. Orndorff v. Lockhart, 906 F.2d 1230 (8th Cir.1990), cert. denied, — U.S. -, 111 S.Ct. 1338, 113 L.Ed.2d 269 (1991). On remand from this court, the district court-held that the error was harmless with respect to the petitioners’ sentences and denied their requests for habeas relief. We affirm in part and reverse and remand, in part.
I.
Petitioners broke into and robbed the house of Don Lehman in Rogers, Arkansas. In addition to stealing several hundred dollars and some guns,, one or more of the petitioners shot' and killed Don Lehman and threatened to kill his wife and daughter, who were in the house at the time.
At trial, Don Lehman’s daughter, Vicki Lehman, testified in great detail about the murder-robbery. In addition to her trial testimony, she gave the police two statements which were recorded and transcribed: one, within hours after the murder and another several days later. Although there are some variations between her trial testimony and her two pretrial statements, there are many similarities in her descriptions of what happened that night.
Vicki Lehman has consistently stated that at approximately 9:45 on the evening of January 8, 1981, she heard the front doorbell ring and got out of bed to go to the door. Vicki met'her father in the hall on the way to the door and followed closely behind him. When her father unlocked the door, four men wearing ski masks forced their way into the house and attacked him. Vicki was held at gunpoint by one of the men while two or three others attempted to subdue her father who was struggling. Vicki was ordered to kneel down, face the wall, and put her head in her hands. She could sense that the fight had moved down the hall and into the bedroom. Vicki then heard two gunshots. Moments later, one of the men grabbed her by the back of the hair and led her through the house, searching for money and other valuables.
Vicki has consistently described the four attackers as follows: (1) the tallest man who wore wire rim glasses that were visible through the ski mask, later identified as petitioner Richley; (2) the heavy or biggest man, later identified as petitioner Holmes; (3) the short or smallest man, later identified as petitioner Orndorff;- and (4) the medium-sized man who removed his mask during the robbery, later identified as petitioner Clines. She got a very good look at the man who removed his mask, and within hours after the murder helped an investigator create a composite drawing of that man. Several days later she positively identified petitioner Clines in a line-up.
■ Vicki has also consistently described the sequence of events as she was led through the house. After she heard the gunshots, the tallest man with the glasses grabbed her and demanded to know where some money [1429]*1429was. This tall intruder never let go of her until he left and constantly had his gun pointed at her head or back. She led him into her bedroom and produced an envelope containing some money she had received for Christmas. From her bedroom, Vicki was taken back down the hall, upstairs to a loft or office area above the main room, and then back downstairs. She was then led into her parents’ bedroom where she saw her father laying motionless on the bed and her mother cowering in the corner. While in that bedroom, the medium-sized man found several hundred dollars in a drawer. He became excited and removed his mask. Then he and the heavy man began tearing out drawers and throwing clothes about the room searching for more money. The smallest man then suggested that she be allowed to call an ambulance for her father.
Vicki was taken into the kitchen area to use the phone. Before she was able to make the call, however, the medium-sized man grabbed the phone from her and stated she was not going to call anyone. Vicki then stared eye-to-eye with this unmasked man! All four attackers were in the kitchen area and they demanded more money and guns. Vicki directed them to the gun cabinet in the living room. The tallest man, still holding Vicki, ordered two of the attackers, the smallest and medium-sized man, to get in the car and ordered the heavy man to remove the guns from the cabinet. The heavy man took the guns out to the car and the tallest man followed.
The police arrested all four petitioners within several days of the murder. After a joint trial, petitioners were convicted of armed robbery and capital murder. The death penalty was imposed on all four.1 Petitioners exhausted their state remedies and filed petitions for habeas corpus relief in federal court. Petitioners raised several claims in their habeas petitions, but only one was found to have merit.
After petitioners had appealed to the Arkansas Supreme Court, they discovered that Vicki Lehman had been hypnotized prior to trial in an attempt to enhance her memory. The .prosecutor had arranged for this hypnosis, but failed to inform the defendants or the trial court. In their habeas petitions, petitioners claimed they were denied their sixth amendment right to confront the witnesses against them because they were denied the opportunity to cross-examine Vicki Lehman abqut the hypnosis.2 The district court found that the prosecutor’s failure to inform the defendants about this hypnosis was constitutional error. The court further found that this error was harmless with respect to the petitioners’ convictions of capital murder, but held that the error was not harmless with respect to the petitioners’ sentences. Orndorff v. Lockhart, 707 F.Supp. 1062 (E.D.Ark.1988).
On appeal from that decision, we affirmed the district court’s holding that the error did not warrant reversal of the convictions because “[t]he evidence of petitioners’ guilt, even without Vicki Lehman’s testimony,' is overwhelming.” Orndoff, 906 F.2d at 1233. We remanded the case, however, with directions for the district court to reconsider the question whether the error was harmless with respect to the petitioners’ death sentences. Id. On remand, the district court found the error harmless and denied petitioners’ request for habeas relief. Petitioners appeal from that' decision.
II.
We begin with the premise that when reviewing a constitutional violation for harmless error, courts traditionally required “the beneficiary of a constitutional error to prove beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained.” Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 828, 17 L.Ed.2d 705 [1430]*1430(1967). The Supreme Court has recently altered this premise, holding that the correct harmless error standard in most habeas corpus cases is "whether the error `had substantial and injurious effect or influence in determining the jury's verdict.'" Brecht v. Abrahamsom, - U.S. -, -, 113 S.Ct. 1710, 1722, 123 L.Ed.2d 353 (1993) (quoting Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 1253, 90 L.Ed. 1557 (1946)).3 This court has recently noted the question whether the Brecht standard applies if the state courts never considered the harmle~s error issue because they found no error. Hoversten v. Iowa, 998 F.2d 614; No. 92-2402 (8th Cir.1993). In Hoversten, however, this court declined to answer that question because the error was not harmless under either standard. Id. 998 F.2d at 617. We now reach the unresolved question.
Brecht reached the Supreme Court after two state appellate courts, a federal district court, and a federal court of appeals had reviewed the error under the Chapman harmless error standard. Petitioners argue that the Court declined to apply the Chapman harmless error standard on collateral review because to do so would undermine "the finality of convictions that have survived direct review within the state court system." Brecht, - U.S. at -, 113 S.Ct. at 1720 (citations omitted). The Court noted that "[s]tate courts are fully qualified to identify constitutional error and evaluate its prejudicial effect on the trial process under Chapman," and that "it scarcely seems logical to require federal habeas courts to engage in the identical approach to harmless-error review that Chapman requires state courts to engage in on direct review." Id. at -, 113 S.Ct. at 1721.
We believe the Brecht rule is based largely on the notion that because the state courts can properly apply the Chapman harmless error standard on direct review, the federal habeas courts need only review those decisions under the Kotteakos harmless error standard. In the case before us, however, the state courts did not have the opportunity to review the error at all, and the federal habeas court was the first court to review the constitutional error under the Chapman harmless error standard. As a result, we find that rule announced in B'recht does not apply and that the Chapman harmless error standard is the appropriate test in this case.
We therefore must determine whether the errorin this case "was harmless beyond a reasonable doubt." Chapman, 386 U.S. at 24, 87 S.Ct. at 828. More specifically, in the context of this confrontation clause violation, "{t]he correct inquiry is whether, assuming that the damaging potential of the cross-examination were fully realized, a reviewing court might nonetheless say that the error was harmless beyond a reasonable doubt." Delaware v. Van Arsdall, 475 U.S. 673, 684, 106 S.Ct. 1431, 1438, 89 L.Ed.2d 674 (1986) (listing five factors a reviewing court should evaluate in making its decision).
In remanding this case, we instructed the district court to conduct its inquiry using the harmless error analysis set forth in Van Arsdali. We noted that the court should compare the pre-hypnotic statements to the trial testimony "with an eye toward sorting out those things abouf which Vicki Lehman could testify without the aid of hypnosis (that is, from actual memory) from those things about which she could testify only after being hypnotized." Orndorff, 906 F.2d at 1232. We further stated that "[u]nless this sorting process reveals significant differences, the error should be deemed harmless." Id. (footnote omitted). The district court found there to be a "remarkable consistency" in Vicki Lehman's recitations of the murder-robbery and therefore concluded the error was harmless.
After the district court made its ruling on remand, the Supreme Court decided Yates v. Evatt, - U.S. -, 111 S.Ct. 1884, 114 L.Ed.2d 432 (1991). Petitioners argue that Yates changes the legal test announced in Van Arsdall and therefore the case must be remanded because the district court's legal [1431]*1431analysis is flawed.4 Respondent Lockhart argues that Yates deals with harmless error in the context of an improper presumption and therefore is not relevant to this confrontation clause harmless error analysis. We find that Yates is consistent with Van Arsdall and does not require us to remand.
In Yates, it was determined that the trial court’s jury instructions created certain presumptions which improperly shifted the burden of proof from the prosecution to the defendant. In analyzing the question whether this unconstitutional burden-shifting jury instruction was harmless, the Court employed a two-step approach. In essence, a reviewing court must first establish “what evidence the jury actually considered in reaching its verdict.” — U.S. at -, 111 S.Ct. at 1893. The purpose of this first step is to determine what evidence may have been considered as tending to prove the improperly presumed fact. The second step is to “weigh the probative force of that evidence [determined in the first step] as against the probative force of the presumption standing alone.” Id. The Court concluded that the proper test is “whether the jury actually rested its verdict on evidence establishing the presumed fact beyond a reasonable doubt, independently of the presumption.” Id.
Petitioners argue that the methodology used in Yates should be “extrapolated” to the present harmless error analysis of this confrontation clause violation. They argue that Yates requires a reviewing court to first determine what evidence the jury considered on the issue of the death sentence. According to petitioners, the court must then determine if there are any differences in the pre- and post-hypnotic testimony of Vicki’ Lehman, and finally determine whether the verdict of death actually rested on the hypnotically enhanced testimony. We agree with petitioners that Yates is instructive of our analysis, but we do not agree that Yates changes the Van Arsdall analysis.
In analyzing a violation of the confrontation clause for harmless error, the Court in Van Arsdall stated that:
Whether such an error is harmless in a particular case depends upon a host of factors, all readily accessible to reviewing courts. These factors include the importance of the witness’ testimony in the prosecution’s case, whether the testimony was cumulative, the presence or absence of corroborating or contradicting testimony of the witness on material points, the extent of cross-examination otherwise permitted, and, of course, the overall strength of the prosecution’s ease.'
475 U.S. at 684, 106 S.Ct. at 1438. The fifth factor described in Van Arsdall (the overall strength of the prosecution’s case) requires us to examine the evidence presented by the prosecution’which the jury may have considered in deciding to impose the death sentence. In Lufkins v. Leapley, 965 F.2d 1477 (8th Cir.1992), we analyzed a confrontation clause violation for harmless error and cited Yates for the proposition that “to determine whether the [error] was harmless beyond a reasonable doubt, we must examine the other evidence introduced at trial and determine whether it appears ‘beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained.’ ” Id. at 1481.
In sum, our analysis under Yates, Van Arsdall, and Lufkins must proceed as follows. We must first compare the pre- and post-hypnotic statements to determine if any “significant differences” exist. If no significant differences are found, the error is deemed to be harmless. If, however, significant differences do exist, we must determine if the variations nevertheless are harmless. To accomplish that task, we assume that Vicki Lehman had been cross-examined about the hypnosis, and we also assume that the damaging potential of that cross-examination was fully realized. See Van Arsdall, 475 U.S. at 684, 106 S.Ct. at 1438. That is, we assume the jury was made aware of any [1432]*1432changes in Vicki Lehman’s story after the hypnosis, as well as the possibility of “confabulation,” “suggestibility,” and “memory-hardening” that is associated with hypnosis.5 After making these assumptions, we must determine if we can still say that the error was harmless beyond a reasonable doubt. To determine whether the defendants’ .inability to cross-examine Vicki Lehman about the hypnosis did not contribute to the jury’s imposition of the death penalty, we must examine the evidence introduced at trial which supported the verdict, as well as the other factors listed in Van Arsdall.
III.
We review the district court’s harmless error analysis de novo, as a mixed question of law and fact. See, e.g., Gunn v. Newsome, 881 F.2d 949, 964 (11th Cir.1989) (en banc); Graham v. Wilson, 828 F.2d 666, 669-60 (10th Cir.1987). We therefore begin our analysis, as did the district court, with a comparison of Vicki Lehman’s pre- and posthypnotic statements. On appeal, petitioners claim Vicki Lehman made 134 post-hypnotic statements which contain significant variations from her two pre-hypnotic statements. They contend 83 of the statements are “new” and 51 of the statements are “changed” from her pre-hypnotic statements.6
Petitioners emphatically claim that Vicki Lehman has been inconsistent in remembering the number of guns she saw.7 Petitioners argue that Vicki’s pre-hypnotic statements indicate that at different times she saw either two or three guns, while at trial she testified that all four petitioners had guns.
Vicki Lehman has stated that she remembered seeing the petitioners with guns at two distinct times during the robbery. The first instance was immediately after the petitioners forced their way into the house. Vicki stated in her second pre-hypnotic statement that when the intruders forced their way in the door, three of the men pointed their guns at her for an instant while the fourth man attacked her father. Of those three men with guns, one then turned to the fight with her father, the tallest man grabbed her and held her at gunpoint, and the smallest man stood nearby with his gun. In her first statement, Vicki stated that two men immediately began wrestling with her father and the other two had their guns on her. In any event, Vicki has never wavered from her statement that petitioner Richley, the tallest man, had a gun. She saw the gun pointed at her and felt it pressed to her head and back throughout most of the ordeal. In addition, she has stated twice (once in the first pre-hypnotic statement and once in the second) that petitioner Orndorff, the smallest man, had a gun when he first entered the house.
The second instance Vicki remembered seeing guns was later that night when all four petitioners were with her in the kitchen. She recounted this event consistently in both of her pre-hypnosis statements. Vicki was taken to the kitchen from her parents’ bedroom to call an ambulance, but petitioner Clines intervened and ripped the phone from the wall. Moments later, three of the petitioners stood in front of her with their guns. At trial, Vicki testified that she believed at that moment all four men had guns in their hands, one at her back and three in front of her. She testified later on cross-examination, however, that it was possible that there were only two guns pointed at her, but “it seem[ed] like” there were three guns in front of her.
In sorting through these statements, we conclude, as did the district court, that there is no, matérial or significant difference between Vicki Lehman’s pre- and post-hypnotic statements concerning the number of guns involved in this crime. She indicated before and after hypnosis that at least three of the [1433]*1433petitioners, and possibly all four, carried guns.
Petitioners argue that Vicki Lehman’s pre-hypnotic statements indicate that only three of the petitioners were in the bedroom at the time Don Lehman was murdered, but that her trial testimony places all four in the bedroom when the shots were fired.8 In her first pre-hypnotic statement, Vicki clearly indicated that the tallest man, Richley, stayed with her in the hall while the others were in the bedroom. In her second pre-hypnotic statement, however, Vicki was less sure that the tall man stayed with her. She indicated that she had her head down on the floor and that he could have gone into the bedroom with the others. Moreover, her second pre-hypnotic statement indicates that she was not positive that the man who originally grabbed her was the same man that grabbed her after the shots were fired. At trial, Vicki admitted on cross-examination that it was possible that not all of the petitioners were in the bedroom at the time she heard the shots fired.
Perhaps the most dramatic addition to Vicki Lehman’s pre-hypnotic statements is her trial testimony about the behavior of petitioner Orndorff while inside the house.9 At trial, she described Orndorff as “wirey acting” and “squirrely-acting.” She testified that Orndorff liked to play with the little toys located on a bar in the main room and specifically testified that he liked, to pull down the lever on the toy slot machine. There is no mention of this behavior in her pre-hypnotic statements. She also stated that “this little guy, the guy who was darting around all night, he kind of ran through there and swung on a chandelier and screamed real loud like he was happy or something.” This incident allegedly occurred within minutes of Don Lehman’s murder as Vicki was being taken into the kitchen to use the phone. The only pre-hypnosis mention of Orndorffs behavior in this regard was in response to a question by the police about the broken glass in the dining room. Vicki replied, “This happened later right before they left. He [referring to Orndorff] just went through there and grabbed it.” At trial Vicki also described an incident that occurred in her parents’ bedroom while her father lay dying on the bed. She stated that the
smaller guy was kind of kicking around through the — through the clothes ... and somebody commented that they’d better get out of there, that the neighbors had probably heard the guns and the smallest guy looked out the window. I can remember seeing him look out the window and he sáid something, I’m not real sure what he said, something about, ‘The neighbors will see us,’ or something.
This incident is not mentioned in either of the pre-hypnotic statements.
Petitioners also claim that Vicki Lehman changed her statements regarding petitioner Holmes’ involvement. At trial, Vicki testified that after the phone was ripped off the wall, all four petitioners were .with her in the kitchen “and they said, ‘Where’s any more money or where’s some jewelry.’ I can remember the big guy going, ‘Guns. Guns. We know you have guns.’ ”10 Petitioners claim this testimony is significantly different from her first pre-hypnotic statement in which she indicated that all' three petitioners standing in front of her at that time demanded the guns, and from her second pre-hyp-notie statement in which she said that the unmasked man, petitioner Clines, demanded the guns. Although her trial testimony is inconsistent with her second pre-hypnotic statement, it is not significantly different from her first pre-hypnotic statement in which she indicated that petitioner Holmes was one of the three men, standing in front of her demanding guns.
At 'trial, Vicki testified that petitioner Holmes was the first one on her father when the petitioners forced their way through the door.11 Petitioners claim this is a significant change from her pre-hypnotic statements. In both the pre-hypnotic statements, however, petitioners Holmes and Clines were im[1434]*1434mediately on her father as soon as the door opened and petitioners Richley and Orndorff stayed by her. The fact that at trial Vicki identified Holmes as being the first of two attackers who immediately pounced on her father is insignificant.
Some of the statements identified by the petitioners are not of any real consequence to the jury’s determination of the sentence for various reasons. Some deal with events before the intruders broke into the house12 or after the intruders left.13 Several of the statements contain physical descriptions of the intruders14 and, in particular, their shoes.15 Others are, descriptions of events that occurred during the robbery and are simply not new or changed.16
We can describe the remaining new or changed statements as adding detail, col- or, vividness, and perhaps credibility to Vicki Lehman’s pre-hypnotic statements. Some of these statements describe in greater detail what occurred immediately after the petitioners forced their way into the house.17 Others add detail to the actual fight that ensued.18 These latter details are indeed new statements because Vicki’s pre-hypnotic statements indicated that she only “sensed” that the fight had moved into the bedroom. At .trial she described.her father getting away from the petitioners and being chased down the hall. Many of the statements describe various details as she walked from room to room with the intruders.19 Several statements deal with Vicki’s own thought processes or feelings during the robbery.20 Other statements describe her mother’s actions.21 When viewed as a whole, these variations may have enhanced Vicki Lehman’s credibility and therefore could have affected the jury’s decision. However, as we noted in our prior decision, the “error is not harmful simply because the jury might have found her testimony less credible had defense counsel been able to probe the issue of her hypnosis on cross-examination.” Orndorff, 906 F.2d at 1232.
The district court found that no significant differences existed between the pre- and post-hypnotic statements and therefore determined the error to be harmless without discussing the analysis set forth in Van Arsdall. We believe there to be enough variation, however, to warrant an examination of the differences in light of the Van Arsdall factors and we may make our own harmless error determination without remanding to the district court. See Yates, — U.S. at -, 111 S.Ct. at 1896 (“Although our usual practice in pases like these is to reverse and remand for a new determination under the correct standard, we have the authority to make our own assessment of the harmlessness of a constitutional error in the first instance.”); see also Lufkins, 965 F.2d at 1481 (holding that in certain circumstances, a reviewing court may conduct a harmless error analysis sua sponte).
The first four factors described in Van Arsdall clearly weigh in favor of petitioners. First, Vicki Lehman’s trial testimony was very important to the prosecution’s case; the prosecutor repeatedly said so to the jury. See Orndorff, 707 F.Supp. at 1068. Second, her testimony was certainly not cumulative since she was the only occurrence witness, other than her mother who saw and remembered very little, to testify for the state. Third, the only person to either corroborate or contradict her testimony on any material issues was petitioner Holmes. And fourth, the defense attorneys understandably cross-examined Vicki rather gingerly about any [1435]*1435variations that they discovered in her trial testimony.
This brings us to our discussion of the fifth factor, the overall strength of the prosecution’s case. Petitioners argue that they are entitled to an individualized evaluation and we agree. In this case, the prosecution’s case against each defendant for the death penalty was different and we believe Yates and Van Arsdall require us to evaluate the case against each petitioner individually. For each petitioner, we will consider the variations in Vicki Lehman’s testimony (that is, those things about which she could testify only after hypnosis) together with the evidence of aggravating and mitigating circumstances, and determine if we can nonetheless say that the error was harmless beyond a reasonable doubt.
We find the prosecution’s case against petitioner Richley for imposition of the death penalty was strong. The jury found that all three applicable statutory aggravating circumstances existed22 and that no mitigating circumstances existed. Vicki Lehman consistently described him as the leader of the group, and consistently stated that he threatened to kill her. At the penalty phase of the trial, the prosecution presented evidence that Richley had planned and executed a series of armed robberies in the months prior to the Lehman murder. The only variation in Vicki Lehman’s statements which may have affected Richley’s sentence was her testimony that all four petitioners were. in the bedroom when the shots were fired. Although Vicki’s first pre-hypnotic statement indicated that Richley was not in the room at the time the shots were fired, her second pre-hypnotic statement indicated that Richley could have been in the room. We can say, however, beyond a reasonable doubt that the error in this case did not contribute to petitioner Richley’s death sentence.
We find that the .prosecution’s case against petitioner Holmes for imposition of the death penalty was strong. More importantly, we find that few if any of the variations in Vicki Lehman’s statements were ma-, terial to the prosecution’s case against Holmes. The variations concerning Holmes involved Vicki stating that Holmes was the first man on her father and that Holmes was demanding the guns. As we discussed above, these statements are not really variations a,t all, and to the extent they are different, the differences are insignificant. The prosecution presented evidence which linked Holmes to’ several of the robberies allegedly committed by Richley. The jury found that all three statutory aggravating circumstances existed, and that no mitigating circumstances existed. The error in this ease did not contribute to petitioner Holmes’ death sentence beyond a reasonable doubt.
We also find that the prosecution’s case against petitioner Clines for imposition of the death penalty was strong and that few if any of the variations in Vicki Lehman’s testimony were material to the prosecution’s case against Clines. Vicki’s pre-hypnotic statements consistently identified Clines as one of the two men immediately upon her father and consistently placed him in the bedroom at the time of the murder. Moreover, Vicki consistently described Clines as the man who ripped the phone from the wall and prevented her from calling an ambulance for her father. The jury found all three aggravating circumstances existed and unanimously found that mitigating circumstances did exist by virtue of Clines’ electrocution23 and his excessive drinking. One or more of the jurors found that Clines suffered from a [1436]*1436personality change after the electrocution. Because there are no significant variations in Vicki Lehman’s testimony which may have affected the jury’s decision to sentence Clines to death, we can say beyond a reasonable doubt that the error in this case did not contribute to petitioner Richley’s death sentence.
We find that the prosecution’s case against petitioner Orndorff for imposition of, the death penalty was the weakest in this case. We also find that the variations in Vicki Lehman’s testimony about Orndorffs behavior were the most significant. The jury found that only two of the three aggravating circumstances existed (the jury did not find that Orndorff created a great risk of death to a person other than the victim) and one or more of the jurors found that two mitigating circumstances existed (that Orndorff was acting under unusual pressures or influences or under the domination of another person and that Orndorff was an accomplice and his participation was relatively minor). We cannot say beyond a reasonable doubt that the variations in Vicki Lehman’s testimony did not contribute to the jury’s decision to sentence Orndorff to death. Without Vicki Lehman’s descriptions of Orndorff playing with toys, swinging from the chandelier,- and looking out the window for the neighbors, the jury may have sentenced Orndorff to life imprisonment rather than death.
We therefore affirm the district court’s denial of habeas relief to petitioners Richley, Holmes, and Clines. We reverse the denial of habeas relief to petitioner Orndorff and, in accordance with our previous decision, Orndorff, 906 F.2d at 1233, we remand the case to the district court with instructions that the court’s order granting habeas relief should give the state the option of conducting a new sentencing proceeding or reducing the sentence to life without parole.