Argued: April 29, 1998
Decided and Filed: November 16, 1998 Pursuant to Sixth Circuit Rule 24.
Appeal from the United States District Court for the Middle District of Tennessee at Nashville. No. 92-00180 — John T. Nixon, District Judge.
ARGUED: Gordon W. Smith, OFFICE OF THE ATTORNEY GENERAL, CRIMINAL JUSTICE DIVISION, Nashville, Tennessee, for Appellant.
Henry A. Martin, FEDERAL PUBLIC DEFENDER'S OFFICE, Nashville, Tennessee, for Appellee.
ON BRIEF: Gordon W. Smith, John Knox Walkup, Glenn R. Pruden, OFFICE OF THE ATTORNEY GENERAL, CRIMINAL JUSTICE DIVISION, Nashville, Tennessee, Michael E. Moore, John H. Baker, III, OFFICE OF THE ATTORNEY GENERAL, Nashville, Tennessee, for Appellant.
James H. Walker, Waller, Lansden, Dortch Davis, Nashville, TN, Henry A. Martin, Paul R. Bottei, FEDERAL PUBLIC DEFENDER'S OFFICE, Nashville, Tennessee, for Appellee.
Before: BOGGS, NORRIS, and MOORE, Circuit Judges.
BOGGS, J., delivered the opinion of the court, in which NORRIS, J., joined. MOORE, J. (pp. 61-65), delivered a separate dissenting opinion.
BOGGS, Circuit Judge.
Ricky Bell, a warden for the State of Tennessee, appeals from the district court's grant to Robert Coe of a writ of habeas corpus, which vacated his murder conviction and death sentence. Coe cross-appeals from the district court's denial of habeas corpus on several other grounds he raised below. On the state's appeal, we reverse on all counts. On Coe's cross-appeal, we affirm on all counts.
Cary Medlin, an eight-year-old girl, was murdered in 1979. She disappeared at around 5:30 p.m. on September 1 while riding bicycles with her eight-year-old step-brother, Michael. She was seen talking to a man in a brown car, and then getting into the car. At trial, Michael identified the driver of the car as Coe.
Medlin's body was discovered the next day at about 2:00 p.m. She had been raped, sodomized, strangled, and stabbed in the neck, in that order. The list of suspects was initially quite long, and at least one other suspect was taken into custody, though he was released for lack of evidence. The search for the killer soon focused on Coe, who was arrested on September 4th while waiting to take a bus to Georgia under an assumed name.
Shortly after arriving at the police station, Coe confessed. He was interrogated, and he offered details of the crime. On the 5th, Coe led officers on a trek to retrace his steps in committing the murder. He pointed out a house where a witness had seen him and his victim, though the witness could identify only the victim, Medlin, as having been in a car that drove by.
On the 7th, Coe gave a statement in which he gave the following account of the events leading to Medlin's death. Coe said that he took Medlin to the spot where her body was eventually found. He exposed himself to her, fondled her, masturbated in front of her, and got on top of her, though he was vague as to what the latter action entailed. At that point, Medlin told Coe that Jesus loved him, a statement that he said caused him to snap. He tried to choke her, and when that did not work, he stabbed her and watched her bleed to death. He then disposed of her body, her shoes, and the knife.
Other evidence incriminated Coe. He apparently came home the night of the 1st and told his family that he had stabbed a state trooper in the throat. He had his wife dye his hair a darker color. The day he was arrested, Coe had traded in his silver and brown car for a blue one.
There was relatively little physical evidence. Coe's car yielded no evidence of a sexual assault. No hairs or fingerprints were found and used against Coe. However, the police did find fecal matter beneath his foreskin, and stains on the front inside of his pants that matched stains found on Medlin's underpants (both were reddish-brown and contained potato starch).
Coe had a history of mental illness. His childhood, according to one expert witness, was "like something out of Erskine Caldwell." His father sexually abused Coe, and forced him to watch while he also sexually abused Coe's sisters. Although several experts testified that Coe was not legally insane at the time of the murder, others testified that he was psychotic, schizophrenic, intoxicated, and under the influence of drugs. In 1975, Coe had been found incompetent to stand trial in Florida after he attempted to rape and then stab a forty-year-old woman.
In 1981, a Tennessee jury convicted Coe of first-degree murder, aggravated rape, and aggravated kidnapping. He was sentenced to death on the first charge, and life imprisonment on the other two. The Tennessee Supreme Court affirmed the conviction and sentence, State v. Coe, 655 S.W.2d 903 (Tenn. 1983), and the United States Supreme Court denied certiorari, Coe v. Tennessee, 464 U.S. 1063 (1984).
Coe first applied for post-conviction relief in state court in 1984. The trial court denied relief after an evidentiary hearing in 1986, and the court of appeals affirmed. Coe v. State, No. C.C.A. 15, 1986 WL 14453 (Tenn.Crim.App. Dec. 23, 1986). The Tennessee Supreme Court denied Coe's request for permission to appeal because he did not timely file for it (JA 227).
In 1987, Coe filed his first petition for habeas corpus relief in federal court. The court dismissed the petition without prejudice in 1989, because Coe had not exhausted his state remedies.
Coe filed his second motion for state post-conviction relief in 1989. It was dismissed and the court of appeals again affirmed. Coe v. State, No. 138, 1991 WL 2873 (Tenn.Crim.App. Jan. 16, 1991). The Tennessee Supreme Court denied permission to appeal, this time on the merits (JA 316).
Coe filed the present federal habeas petition in 1992. During the pendency of this case, he filed a third motion for state post-conviction relief, which was denied. The court of appeals affirmed. Coe v. State, No. 02C01-9606-CR-00200, 1997 WL 88917 (Tenn.Crim.App. Mar. 4, 1997). The Tennessee Supreme Court granted permission to appeal in December 1997. Ibid.
Coe amended his federal petition in 1995 and 1996. The latter amendment included only part of what Coe wanted to add, as discussed further below, see infra, at 29-31. The district court disposed of some of Coe's claims in April 1996, when it granted partial summary judgment in favor of the state. It disposed of the rest in December when, after an evidentiary hearing, it granted Coe relief on five of his claims, and denied relief on all of the others. Both parties timely appealed.
All of the grounds on which the district court granted habeas relief had to do with jury instructions. To warrant habeas relief, the jury instructions must have been so infirm that they rendered the entire trial fundamentally unfair. An ambiguous, potentially erroneous instruction violates the Constitution only if there is a reasonable likelihood that the jury has applied the instruction improperly. Estelle v. McGuire, 502 U.S. 62, 72 (1991); Austin v. Bell, 126 F.3d 843, 846 (6th Cir. 1997), cert. denied, 118 S.Ct. 1547 (1998). For capital sentencing factors, there are additional considerations, which we discuss below.
A. Guilt Phase Instructions
The district court granted habeas and vacated Coe's conviction on two grounds from the guilt phase: reasonable doubt instructions and malice instructions.
1. Reasonable doubt
The district court ruled that the following instruction on reasonable doubt at the guilt phase was impermissible, and reversed all of Coe's convictions:
Reasonable doubt is that doubt engendered by an investigation of all the proof in the case and an inability, after such investigation, to let the mind rest easily upon the certainty of guilt. Reasonable doubt does not mean a doubt that may arise from possibility. Absolute certainty of guilt is not demanded by the law to convict of any criminal charge, but moral certainty is required and this certainty is required as to every proposition of proof requisite to constitute the offense.
(emphasis added). A functionally equivalent instruction was given at the sentencing stage.
Subsequent to the district court's decision, we approved the identical instruction in Austin, 126 F.3d at 846-47. Coe concedes this and offers no reason why we should overrule ourselves, and we shall not.
The district court found that two parts of the jury instructions on malice were constitutionally flawed, and vacated Coe's murder conviction.
We do not reach the merits of these challenges, because Coe's claim is procedurally barred from being considered here. Procedural bar applies when a state prisoner defaults his federal claims in state court pursuant to an independent and adequate state procedural rule. Simpson v. Sparkman, 94 F.3d 199, 202 (6th Cir. 1996). There are exceptions to this rule when the prisoner can demonstrate cause for the default and actual prejudice as a result of the alleged violation of federal law, or can demonstrate that failure to consider the claims will result in a fundamental miscarriage of justice. Ibid. When a state-court judgment rested primarily on federal law or was interwoven with federal law, the bar applies only if the state court clearly and expressly stated that its judgment rested on a state procedural bar. Ibid.
Coe did not raise his malice instruction argument at trial, on direct appeal, or in his first state post-conviction motion. He did so in his second effort at state post-conviction relief, but the state trial court held that the claim was procedurally barred because it had not been raised before. The court of appeals agreed, holding that Coe's failure to raise the claim was not excusable based on the argument that its basis was new. The claim was based on Yates v. Aiken, 484 U.S. 211 (1988), but that case held only that Francis v. Franklin, 471 U.S. 307 (1985), applied retroactively. Coe could have asserted a Francis claim in his first state motion for post-conviction relief, which was not decided until 1986. Because he had not done so, the court said, the claim was barred.
The court also said:
Even if we assume that the instructions on malice were erroneous under Francis, and these issues were not waived in Coe I, any error was clearly harmless under the facts of this case. The facts established both felony murder and common-law murder beyond a reasonable doubt. State v. Coe, 655 S.W.2d 903 (Tenn. 1983). Malice is immaterial to felony murder. State v. McKay, 680 S.W.2d 447 (Tenn. 1984). The facts in this case clearly demonstrate an overwhelming amount of evidence of malice. Ground 7 is without merit.
Coe, 1991 WL 2873, at *6. Coe appears to have challenged the waiver holding in his effort to appeal, but the Tennessee Supreme Court denied Coe permission to appeal because his filing was untimely.
Coe claims that the court of appeals's alternative holding — that he would lose on the merits anyway — means that he is not procedurally barred, because the state courts in fact reached the merits. This argument fails due to the Supreme Court's decision in Harris v. Reed, 489 U.S. 255 (1989). Harris states that a state court need not fear reaching the merits of a federal claim in an alternative holding. By its very definition, the adequate and independent state ground doctrine requires the federal court to honor a state holding that is a sufficient basis for the state court's judgment, even when the state court also relies on federal law.
Id. at 264 n. 10 (citing Fox Film Corp. v. Muller, 296 U.S. 207, 210 (1935)); see also Sochor v. Florida, 504 U.S. 527, 533 (1992). The alternative holding thus does not require us to disregard the state court's finding of procedural bar.
Coe next argues that the court of appeals ruling does not represent a "clear and express" statement of procedural bar. Although it could be clearer and more express, the court's statement that (1) Coe's argument did not "present a cognizable claim" because it was not based on new law; coupled with (2) the self-consciousness of the alternative argument ("even if . . . these issues were not waived in Coe I"), suffices as a clear and express statement. Furthermore, it is worth noting that Coe's appeal to the state supreme court from this ruling discussed only procedural bar, and not the court's holding on the merits.
By way of comparison, the state-court statement in Harris, the case that established the "clear and express" doctrine, was held to be unclear:
In its order, the Appellate Court referred to the "well-settled" principle of Illinois law that "those [issues] which could have been presented [on direct appeal], but were not, are considered waived." The court found that . . . petitioner's ineffective-assistance allegations "could have been raised in [his] direct appeal." The court, however, went on to consider and reject petitioner's ineffective-assistance claim on its merits.
Harris, 489 U.S. at 258 (alterations in Harris) (citations omitted). In other words, the state appellate court in Harris said "X means waiver, and this case has X." Coe's court took things one step further, however, and explicitly and clearly said that Coe had no cognizable claim. There was, therefore, a sufficiently clear and express statement here.
Coe next argues that his violation of state procedural standards was not "knowing and understanding" as required by state statute. That is, he claims that the court of appeals was wrong to hold that his claim was procedurally barred (if it so held) because he did not know that omitting the claim from his first state motion for post-conviction relief would prevent him from raising it later. This argument is foreclosed by House v. State, 911 S.W.2d 705, 714 (Tenn. 1995), cert. denied, 517 U.S. 1193 (1996), in which the Tennessee Supreme Court decided to use an objective waiver standard rather than a subjective one.
Although the petitioner in House was represented by counsel in his first post-conviction proceeding (in which he waived certain claims), and Coe proceeded pro se, the holding in House does not turn on this distinction. The court in House takes pains to say that there is no right to counsel in post-conviction proceedings, and so the ineffectiveness of post-conviction counsel does not excuse waiver. Id. at 712. In other words, even if Coe had had ineffective counsel (i.e. the functional equivalent of no counsel at all) he would still be bound by his omissions in his first motion for post-conviction relief. The key is that Coe had a full and fair hearing in his first motion on whatever claims he chose to raise, see id., and he does not allege otherwise here.
Coe offers one more rejoinder. He cites Hathorn v. Lovorn, 457 U.S. 255 (1982) for the proposition that "a state procedural ground is not `adequate' unless the procedural rule is `strictly or regularly followed.'" Id. at 262-63 (quoting Barr v. City of Columbia, 378 U.S. 146, 149 (1964)). Coe says that because Tennessee's waiver practice has not been consistent in cases with similar circumstances, it cannot serve as the basis for a procedural bar. The state responds by citing numerous cases that establish waiver, and it has the better of the argument. Coe's cases are mostly either adverse, or too old to constitute current "strict and regular" practice. The few that remain are isolated and unpublished, and so are also insufficient to defeat an otherwise "strict and regular" practice. See Delbridge v. State, 742 S.W.2d 266 (Tenn. 1987) (not mentioning waiver); Moore v. State, No. 03C01-9212-CR-00445, 1994 WL 17864, at *3 (Tenn.Crim.App. Jan 25, 1994) (holding that record did not establish knowing waiver); Sneed v. State, No. 03C01-9201-CR-00027, 1992 WL 200951 (Tenn.Crim.App. Aug. 21, 1992) (not mentioning waiver); Bates v. State, No. 03C01-9102-CR-00055, 1991 WL 172999 (Tenn.Crim.App. Sept. 10, 1991) (same); Brewer v. State, No. 1179, 1991 WL 21605, at *2-*3 (Tenn.Crim.App. Feb. 22, 1991) (finding that petitioner did not knowingly waive claim); State v. Bounds, No. C.C.A. 88-170-III, 1989 WL 92215, at *1 (Tenn.Crim.App. Aug. 17, 1989) (finding of no waiver, because of intervening case law). Furthermore, even if the cases were not any of these things, Coe's argument proves too much. Under Coe's theory, the state would never be able to begin using a procedural bar doctrine, because it would not be able to wipe the slate clean of any precedent that accrued before the institution of the procedural bar.
We pause to note that, were we to reach the merits, we would be inclined to reverse anyway.
The first problematic instruction on malice read as follows:
If a deadly weapon is handled in a manner so as to make the killing a natural or probable result of such conduct, then that may be considered by you as to the existence of malice sufficient to support a conviction of murder in the second degree unless it is rebutted by other facts and circumstances.
In Houston v. Dutton, 50 F.3d 381, 385 (6th Cir.), cert. denied, 516 U.S. 905 (1995), a case originating in Tennessee, we invalidated the use of a "deadly weapon" instruction. In that case, however, the instruction said that such handling of a deadly weapon "raises a presumption of malice," subject to rebuttal. Ibid (emphasis omitted). In this case, by contrast, the court said that the use of the weapon "may be considered." This distinction is important. In Francis v. Franklin, 471 U.S. 307, 313-14 (1985), the Supreme Court drew a distinction between "mandatory" inferences, which are problematic, and permissive ones, which are not. The language here is unambiguously permissive. See Peterson v. Murray, 904 F.2d 882, 888 (4th Cir.) (deeming similar "may" language permissive), cert. denied, 498 U.S. 992 (1990); United States v. Washington, 819 F.2d 221, 225-26 (9th Cir. 1987) (same).
The district court argued that the "unless" language in the rebuttal clause implied mandatoriness, since there would be no need for a rebuttal unless there was a mandatory presumption. We disagree. The "deadly weapon" instruction merely sets up a possible inference, and then notes a situation in which even that inference is impermissible. See Elam v. Nix, 951 F.2d 890, 891 (8th Cir. 1991) (reaching same conclusion); cf. United States v. Reeves, 594 F.2d 536, 541 (6th Cir.) (disapproving of "unless" language with less clearly permissive inference), cert. denied, 442 U.S. 946 (1979).
The second challenged instruction reads as follows:
Implied malice may be found to exist where the wrongdoer did not intend to slay the person killed but death resulted from a consciously unlawful act done intentionally and with knowledge on the wrongdoer's part that the act was directly perilous to human life. In this event there is implied such a high degree of conscious and willful recklessness as to amount to that malignity of heart constituting malice.
According to the district court, this instruction requires a finding of malice when the other elements of murder are found; once the jury finds that the defendant acted intentionally and caused the victim's death, the instruction leaves no alternative but to find malice, and so the prosecution is relieved of its burden of proving malice.
The inference in this instruction ("may be found") appears to us to be a permissive one. The several cases cited by the district court in finding this instruction impermissible all involve clearly mandatory language, and none of them contain the "directly perilous to human life" language present here. See Yates v. Evatt, 500 U.S. 391, 401-02 (1991) (disapproving of an instruction that: "`[m]alice is implied or presumed' from the `willful, deliberate, and intentional doing of an unlawful act. . . .'"); Mullaney v. Wilbur, 421 U.S. 684, 686-87 (1975) (disapproving of standard that malice is established by unlawful intentional act, unless heat of passion or provocation were established too); Alexander v. Foltz, 838 F.2d 140, 146 (6th Cir.) (quoting People v. Richardson, 293 N.W.2d 332, 340 (Mich. 1980)) (Ryan, J.) ("The necessary factual element of malice may be permissibly inferred from the facts and circumstances of the killing, but it can never be established as a matter of law by proof of other facts."), cert. denied, 486 U.S. 1033 (1988).
Our certitude is tempered by the next sentence (not considered by the district court), but that sentence appears simply to explain the permissive inference, rather than convert it into a mandatory one. In any event, however, we do not reach the merits on this question, because Coe's claim is procedurally barred.
B. Sentencing Phase
We have previously had occasion to explore the nature of the Tennessee death penalty process:
Tennessee is a "weighing" state — that is, the jury determines whether any aggravating circumstances have been established beyond a reasonable doubt by the State and then balances this against any mitigating circumstances found by the individual jurors. If the jury unanimously finds that the aggravators outweigh the mitigators, death must be imposed.
Houston v. Dutton, 50 F.3d 381, 387 (6th Cir.), cert. denied, 516 U.S. 905 (1995) (emphasis omitted).
1. "Heinous, atrocious, or cruel" instruction
At the sentencing phase, the jury found four aggravating factors that applied and which, on the whole, were not outweighed by mitigating factors. These can be summarized as:
(1) The victim was under 12 and the defendant over 18;
(2) The murder was especially heinous, atrocious, or cruel and involved torture;
(3) The murder was committed for the purpose of avoiding prosecution; and
(4) The murder was committed while the defendant was engaged in committing and fleeing after committing aggravated rape and aggravated kidnapping.
At issue on appeal is the second ground. The district court had defined this factor for the jury as follows:
The murder was especially heinous, atrocious, or cruel in that it involved torture or depravity of mind.
"HEINOUS" means extremely wicked or shockingly evil.
"ATROCIOUS" means outrageously wicked and vile.
"CRUEL" means designed to inflict a high degree of pain, utter indifference to, or enjoyment of, the suffering of others, pitiless.
In holding this factor to be impermissible, the district court cited Houston, and Shell v. Mississippi, 498 U.S. 1 (1990) (per curiam). In Houston, we held that this same Tennessee instruction (minus the appended definitions of the three terms) was unconstitutionally vague. Houston, 50 F.3d at 387. Established Supreme Court precedent had held that simple "heinous, atrocious, or cruel" language is unconstitutionally vague. See Richmond v. Lewis, 506 U.S. 40 (1992); Maynard v. Cartwright, 486 U.S. 356 (1988). We gave no explanation in Houston for why the limitation to "torture and depravity of mind" did not suffice to cure this vagueness problem. Indeed, the Supreme Court suggested that such a limitation might suffice in dicta in Maynard, 486 U.S. at 365. But in Houston, the state conceded that the instruction was vague and we held accordingly.
In Shell, the Supreme Court disapproved of a set of instructions similar to the ones here. The instructions in Shell lacked the "torture and depravity" modifier, but appended individual definitions of "heinous, atrocious, and cruel" that are functionally and virtually equivalent to those used in this case. Shell, 498 U.S. at 2 (Marshall, J., concurring). The Court held that these definitions did not suffice to cure the vagueness problem. Id. at 1; see id. at 2 (Marshall, J., concurring).
In combination, then, Houston and Shell require us to hold that the instructions on this point in this case were constitutionally infirm.
The state offers two arguments against overturning the sentence on this ground. First, it asserts that this claim is procedurally barred for Coe. Despite the state's assertion, however, Coe did raise the issue in his direct appeal, apparently by incorporating an argument from his motion for a new trial that the potential aggravating circumstances presented to the jury were (federally) unconstitutionally vague. Regardless of whether the state supreme court should not have addressed an issue raised in that manner, it upheld the statute on the merits against Coe's challenge. Coe, 655 S.W.2d at 913. The court cited a case that upheld the state statute, State v. Austin, 618 S.W.2d 738, 742 (Tenn.), cert. denied, 454 U.S. 1128 (1981). That case relied on an earlier case, State v. Dicks, 615 S.W.2d 126, 131 (Tenn.), cert. denied, 454 U.S. 933 (1981), that approved the "heinous, atrocious, or cruel . . . torture or depravity of mind" instruction on the purported grounds of a United States Supreme Court decision (Proffitt v. Florida, 428 U.S. 242, 255-56 (1976)). Therefore, the claim is not procedurally barred.
The state next argues that any error stemming from this aggravating factor is harmless. To answer this question, unfortunately, we must venture into a thicket; it is unclear if we may engage in harmless-error analysis when dealing with an infirm aggravating factor, or if instead this is a matter reserved for the state trial and appellate courts. We join the four other circuits that have squarely addressed this question and hold that we are indeed permitted to perform a harmless-error analysis here. See Billiot v. Puckett, 135 F.3d 311 (5th Cir. 1998), cert. denied, ___ S.Ct. ___, 1998 WL 396478 (Nov. 2, 1998); Davis v. Executive Dir. of Dep't of Correc., 100 F.3d 750, 768 n. 18 (10th Cir. 1996), cert. denied, 117 S.Ct. 1703 (1997); Williams v. Clarke, 40 F.3d 1529, 1539-40 (8th Cir. 1994), cert. denied, 514 U.S. 1033 (1995); Smith v. Dixon, 14 F.3d 956, 974-81 (4th Cir.) (en banc), cert. denied, 513 U.S. 841 (1994); see also O'Guinn v. Dutton, 88 F.3d 1409, 1461 (6th Cir. 1996) (en banc) (Batchelder, J., dissenting) (six judges of this court espousing this conclusion in a dissent from a majority opinion that did not reach the merits), cert. denied, 117 S.Ct. 754 (1997).
We must make an initial distinction. This discussion is only an issue in this case because Tennessee is a "weighing" state, in which the jury weighs aggravating factors against mitigating factors. In non-weighing states, the sentencer must find at least one aggravating circumstance to make a convicted murderer eligible for the death penalty. Once the factor is found, the jury weighs the totality of the circumstances. Therefore, if multiple aggravators are found but an appellate court strikes one of them down, the death sentence can stand as it is. This is because the sentencer has still found at least one aggravating factor, and the invalidation of another aggravator does not necessarily change the totality of the circumstances that are considered to arrive at a sentence (though other errors might do so, and could necessitate reversal). See Stringer v. Black, 503 U.S. 222, 229 (1992); Zant v. Stephens, 462 U.S. 862 (1983); see also Tuggle v. Netherland, 516 U.S. 10 (1995). In a weighing state, by contrast, when a court invalidates one of the aggravators, it has removed a mass from one side of the scale. There is no way to know if the jury's analysis — how the aggravating and mitigating circumstances balanced — would have reached the same result even without the invalid factor. Stringer, 503 U.S. at 231-32.
Therefore, whenever an aggravating factor has been invalidated in a weighing state, the sentence must be re-weighed or analyzed for harmless error if the sentence is to be affirmed. Ibid. The question before us, then, is who is permitted to perform such analyses. The state concedes that we may not perform reweighing, but it claims that it is perfectly acceptable for us to engage in harmless-error analysis. We agree that this distinction is warranted. In reweighing, a state court effectively vacates the original sentence and resentences the defendant; this process is hardly appropriate in the course of collateral review by a federal court. In harmless-error analysis, by contrast, a court determines that the original sentence is not constitutionally infirm in the first place, a process that is quite appropriately performed on federal collateral review. Indeed, we would perform a similar analysis if, say, Coe claimed ineffective assistance of counsel based on a failure to raise his vagueness argument on direct appeal. See Smith, 14 F.3d at 976. In that instance, we would have to apply the test of Strickland v. Washington, 466 U.S. 668, 687 (1984), and determine whether Coe had been prejudiced by his counsel's lapse. The analysis we are performing here is not appreciably different.
Coe responds that however sensible the above holding may seem, we are prevented from performing harmless-error analysis by the express language of Supreme Court and Sixth Circuit case law. Although this argument is superficially convincing, we cannot agree.
In Stringer, the Supreme Court held that "[u]se of a vague or imprecise aggravating factor in the weighing process invalidates the sentence and at the very least requires constitutional harmless-error analysis or reweighing in the state judicial system." Stringer, 503 U.S. at 237. The Court subsequently reiterated and clarified this principle, ruling that "[w]here the death sentence has been infected by a vague or otherwise constitutionally invalid aggravating factor, the state appellate court or some other state sentencer must actually perform a new sentencing calculus, if the sentence is to stand." Richmond v. Lewis, 506 U.S. 40, 49 (1992) (emphasis added). Finally, in Houston, we cited Richmond and affirmed the grant of a writ of habeas corpus, because the state court had not recognized the error and thus had not "performed a new sentencing calculus." Houston, 50 F.3d at 387 (quotation marks omitted).
Significantly, though, we did not address the harmless-error question in Houston. Our holding, therefore, applied only to reweighing. Indeed, the portion of Richmond that we quoted (the same portion set forth in the paragraph above) makes this clear: it is only when "the death sentence has been infected by a constitutionally . . . invalid aggravating factor" that state reweighing is required to preserve the verdict. By definition, though, an error that is harmless does not "infect" the sentence and does not require reweighing by the state.
As a final note, the language of Stringer requiring "constitutional harmless-error analysis or reweighing in the state judicial system" is consistent with our conclusion. For the reasons discussed above, the phrase "state judicial system" modifies "reweighing" only, and not "harmless-error analysis." Indeed, the Supreme Court has never held otherwise. Rather, the Court has taught that "[f]ederal intrusions into state criminal trials frustrate both the States' sovereign power to punish offenders and their good-faith attempts to honor constitutional rights." Brecht v. Abrahamson, 507 U.S. 619, 635 (1993) (quoting Engle v. Isaac, 456 U.S. 107, 128 (1982)). Therefore, "before overturning final and presumptively correct state convictions or sentences on habeas review, [we] must assess for harmlessness those errors that are eligible for this review in order to assure that the extraordinary relief provided by the writ is granted only to those `persons whom society has grievously wronged.'" Smith, 14 F.3d at 976 (quoting Brecht, 507 U.S. at 634 (quoting Fay v. Noia, 372 U.S. 391, 441 (1963))).
We turn, therefore, to analyze this error for harmfulness. The question we ask is whether the error "had substantial and injurious effect or influence in determining the jury's verdict." Brecht, 507 U.S. at 623 (quoting Kotteakos v. United States, 328 U.S. 750, 776 (1946)).
The important criterion in a vagueness analysis of an aggravating circumstance is narrowing: "A capital sentencing scheme must, in short, provide a meaningful basis for distinguishing the few cases in which [the penalty] is imposed from the many cases in which it is not." Godfrey v. Georgia, 446 U.S. 420, 427 (1980) (quotation marks omitted; alteration in original).
Our analysis is relatively simple in this case. Even though the aggravator at issue was phrased as "especially heinous, atrocious, or cruel in that it involved torture or depravity of mind," the jury held more narrowly that "the murder was especially heinous, atrocious, or cruel and involved torture." As is evident from the facts recited above, there is more than ample evidence to support such a conclusion.
This distinction — finding torture but not depravity of mind — is significant. The vagueness problem of the "heinous, atrocious, and cruel" ("HAC") instruction is curable with appropriately narrowing language. We have held, of course, that requiring "torture or depravity of mind" does not solve the vagueness problem. Requiring only torture, however, does. See Maynard v. Cartwright, 486 U.S. 356, 364-65 (1988) (implying that "torture" limitation suffices to cure vagueness of HAC); Walton v. Arizona, 497 U.S. 639, 654 (1990) (confirming implication); Duvall v. Reynolds, 139 F.3d 768, 793 (10th Cir. 1998) (holding that "torture of the victim or serious physical abuse" language in the instruction cures vagueness of HAC); cf. Wade v. Calderon, 29 F.3d 1312, 1319-20 (9th Cir. 1994) (holding that intentional torture suffices), cert. denied, 513 U.S. 1120 (1995). In this case, the jury ignored the problematic "depravity" factor and limited its finding to the appropriately narrowing "torture" factor, confirming that finding in a specific, handwritten verdict form. Furthermore, in Maynard and Walton, the narrowing factor was only applied by a reviewing court, not by the jury itself — the fact that we have this evidence from the jury itself confirms our conclusion that the jury's discretion was channeled and narrowed appropriately. The error in this case was harmless, and the district court erred in granting habeas corpus relief on this basis.
Next at issue is the district court's determination that the jury instructions on unanimity in sentencing were unacceptable.
The state first contends that this claim is procedurally barred. When Coe raised this issue in his third state motion for post-conviction relief, the trial court said that he was procedurally barred because he should have raised the issue before, and the court of appeals said that the issue was part of a group of questions that were "waived, previously determined on direct appeal, and/or time barred." The state trial court's statement suffices as a clear statement, and the court of appeals's line-blurring mass affirmance does not change that conclusion. See Ylst v. Nunnemaker, 501 U.S. 797, 803 (1991) ("[W]here, as here, the last reasoned opinion on the claim explicitly imposes a procedural default, we will presume that a later decision rejecting the claim did not silently disregard that bar and consider the merits.").
As with the "heinous" instruction discussed above, however, Coe did raise this issue in his direct appeal, apparently by incorporating it from his motion for a new trial. Also as mentioned above, the state supreme court held that the death-sentence statute was not constitutionally infirm. It is not clear if this holding applies to the unanimity provisions, however, and the cases cited by the state supreme court on direct appeal do not cover unanimity. See State v. Austin, 618 S.W.2d 738, 742 (Tenn.), cert. denied, 454 U.S. 1128 (1981); State v. Dicks, 615 S.W.2d 126 (Tenn.), cert. denied, 454 U.S. 933 (1981). The district court responded to this by citing Tennessee Code § 39-2-205 and State v. Martin, 702 S.W.2d 560, 564 (Tenn. 1985) for the notion that, in capital cases, the state supreme court has to review significant errors, whether or not they were raised by the defendant.
As phrased by the district court, this proposition is too broad, as it would eliminate the entire doctrine of procedural bar in Tennessee in capital cases. See Kornahrens v. Evatt, 66 F.3d 1350, 1362-63 (4th Cir. 1995) (accepting similar reasoning in South Carolina case), cert. denied sub nom. Kornahrens v. Moore, 517 U.S. 1171 (1996). Martin, though, cited § 39-2-205 and reviewed a question that had been discussed but not preserved for review at trial. Martin, 702 S.W.2d at 564. A fortiori, because the issue in this case was not only discussed but formally contested, Martin applies to eliminate the procedural bar problem for Coe. The state court's suggestion of waiver in dismissing Coe's third petition for post-conviction review was only a successive-petition type of waiver; it did not address the issue of whether the question had been raised on direct appeal. Even if it had, furthermore, the third petition is currently pending before the state supreme court.
Therefore, we find that this claim is not procedurally barred, and so we turn to the merits. The jury was told:
If you unanimously determine that at least one statutory aggravating circumstance or . . . circumstances have been proved by the State, beyond a reasonable doubt, and said circumstance or circumstances are not outweighed by any mitigating circumstances, the sentence shall be death. The Jury shall state in writing the statutory aggravating circumstance or . . . circumstances so found, and signify in writing that there were no mitigating circumstances sufficiently substantial to outweigh the [aggravating circumstances].
The jury was then given the form its verdict should take:
(1) We, the Jury, unanimously find the following listed statutory aggravating circumstance or circumstances;
(2) We, the Jury, unanimously find that there are no mitigating circumstances sufficiently substantial to outweigh the [aggravating circumstances] so listed above.
(3)Therefore, we, the Jury, unanimously find that the punishment shall be death.
The alternate result was then provided for and explained:
If you unanimously determine that no statutory aggravating circumstance has been proved by the State beyond a reasonable doubt; or if the Jury unanimously determine that [aggravating circumstances] have been proved by the State beyond a reasonable doubt; but that said [aggravating circumstances] are outweighed by one or more mitigating circumstances, the sentence shall be life imprisonment.
For both the death verdict and the life imprisonment verdict, the jury was told that its verdict must be unanimous.
The district court, relying on two of its own precedents, concluded that these instructions were unacceptable because there was a reasonable probability that the jurors believed that they could consider only those mitigating circumstances that they unanimously agreed were present. See McKoy v. North Carolina, 494 U.S. 433, 439-41, 444 n. 8 (1990); Mills v. Maryland, 486 U.S. 367, 373-75 (1988) (declaring such a requirement unconstitutional). The district court also held that the instructions improperly failed to inform the jury of the consequence of a non-unanimous verdict (i.e., a life sentence).
We first note that this same issue was raised in a recent case, Austin v. Bell, 938 F. Supp. 1308 (M.D. Tenn. 1996), in which the district court reached the same conclusion as it did in this case. See id. at 1320-21. We affirmed Austin on ineffective assistance grounds, and so specifically did not reach the unanimity question, but we noted that although we were not reaching the issue, we had "serious concerns" about the instruction. Austin v. Bell, 126 F.3d 843, 849 (6th Cir. 1997), cert. denied, 118 S.Ct. 1547 (1998). We cited, among other cases, Mills v. Maryland, 486 U.S. 367 (1988).
In Mills, a similar issue was raised; the Supreme Court ruled that the proper inquiry is whether a reasonable jury might have interpreted the instructions in a way that is constitutionally impermissible. Mills, 486 U.S. at 375-76. The relevant portion of the form in Mills read as follows: "Based upon the evidence we unanimously find that each of the following mitigating circumstances which is marked `yes' has been proven to exist by a preponderance of the evidence and each mitigating circumstance marked `no' has not been proven by a preponderance of the evidence." Mills, 486 U.S. at 387 (emphasis omitted). The Supreme Court held this to be impermissible. Id. at 377-84.
We have some concern that it is inappropriate to apply Mills in this case, because it represents a "new rule" under Teague v. Lane, 489 U.S. 288 (1989). Coe's direct appeal process ended in 1984, four years before Mills was decided. Other courts 676, 685-86 (8th Cir. 1995) (finding bar) and Cordova v. Collins, 953 F.2d 167, 173 (5th Cir.) (same), cert. denied, 502 U.S. 1067 (1992), with Williams v. Dixon, 961 F.2d 448, 456 (4th Cir.) (holding that exception to Teague applies), cert. denied, 506 U.S. 991 (1992) and DeShields v. Snyder, 829 F. Supp. 676, 687-88 (D. Del. 1993) (holding that Mills is not a "new rule"). See also Zettlemoyer v. Fulcomer, 923 F.2d 284, 316 n. 3 (3d Cir.) (Sloviter, J., dissenting on other grounds), cert. denied, 502 U.S. 902 (1991). The state has not raised the Teague issue, however, and we need not do so sua sponte. Caspari v. Bohlen, 510 U.S. 383, 389 (1994).
In Kordenbrock v. Scroggy, 919 F.2d 1091, 1108-10, 1120-21 (6th Cir. 1990) (en banc), cert. denied, 499 U.S. 970 (1991), the trial court had given a unanimity instruction with regard to aggravating factors, but not with regard to mitigating ones. We held that the only reasonable reading of the instruction was that, by omission, no unanimity was required as to mitigating factors. Kordenbrock, 919 F.2d at 1121.
We find that the instructions challenged by Coe do not violate Mills. Their language requires unanimity as to the results of the weighing, but this is a far different matter than requiring unanimity as to the presence of a mitigating factor. Nothing in this language could reasonably be taken to require unanimity as to the presence of a mitigating factor. The instructions say clearly and correctly that in order to obtain a unanimous verdict, each juror must conclude that the mitigators do not outweigh the aggravators.
The language certainly is not as directly problematic as that in cases that have followed Mills. See, e.g., McKoy, 494 U.S. at 436 ("Do you unanimously find from the evidence the existence of one or more of the following mitigating circumstances?"); United States ex rel. Kubat v. Thieret, 679 F. Supp. 788, 813 (N.D. Ill. 1988) ("If . . . you unanimously conclude that there is a sufficiently mitigating factor or factors to preclude imposition of the death sentence, you should sign the form which so indicates."), aff'd, 867 F.2d 351 (7th Cir.), cert. denied, 493 U.S. 874 (1989). Those cases did not require any sort of inferential leap to conclude that the jury had to be unanimous before it could even consider a particular mitigating factor; the instruction specifically so stated. In Coe's case, by contrast, it is fairly clear that, as with the first challenged instruction, the unanimity refers to the weighing process and not to the finding of a mitigating factor. Coe argues that the unanimity instruction in the weighing process is improper too, an issue that is addressed below.
We also note that the problematic language did not appear in the section of the instructions on finding mitigating factors. Rather, it appeared in the section on weighing. By contrast, the instructions that precede the section on finding mitigating factors says nothing at all about unanimity. Rather, they say that the jury "shall consider . . . any mitigating circumstance." This is in direct contradistinction to the immediately previous section on aggravating circumstances, where the jury is told that "no death penalty shall be imposed by a Jury but upon an [sic] unanimous finding of the existence of one or more of the statutory aggravating circumstances. . . ." The contrast between the unanimity required for "finding . . . one or more" aggravators and the silence accompanying the instructions on "consider[ing] any" mitigators is precisely what we found dispositive in Kordenbrock. The instructions, therefore, are not improper.
The dissent emphasizes the use of the words "as heretofore indicated" in the instructions on considering mitigation, and argues that those words can easily be understood by the jury to refer to the earlier unanimity discussion on finding aggravating circumstances. However, in context, that construction is very unlikely. The instructions condensed to 12 lines at page 63 of the dissent actually occupy over two pages in the record. There is a whole page omitted at the ellipsis in the block quotation before the crucial language "as heretofore indicated." Also omitted is the key language explaining that it is the statute that provides for the jury to consider "as heretofore indicated" all mitigating factors. The words "as heretofore indicated" are taken directly from § 39-2404(j) of the Tennessee Code, and obviously refer to the whole process of "consideration" that has just been explained, including burden of proof, signing of the verdict form, weighing of mitigating factors against aggravating factors, etc. It is not a plausible construction that those words import a process of unanimity that was required only and specifically of the finding of aggravating circumstances.
Coe naturally emphasizes the less clear-cut weighing language, and valiantly attempts to bring the language within the prohibition of Mills. Coe has noted that an instruction that gets the law right does not necessarily save another, contradictory instruction, since it is impossible to know which of the two contradictory propositions the jury relied upon. See Francis v. Franklin, 471 U.S. 307, 322 (1985). In this case, however, the two passages are not contradictory. One is clear and the other is less clear, to be sure, but they are not incompatible. In Francis, the Court considered whether "taken as a whole [the correct language] might have explained the [law] with sufficient clarity that any ambiguity in the [challenged language] could not have been [misconstrued] by a reasonable juror. . . ." Id. at 318-19. The Supreme Court reminded us, in other words, that we should read the whole of the instructions, and should not isolate and parse text until we find something wrong with it. In Francis, the Court found that the context did not cure the potential for misunderstanding, but in this case we find that it does, to the extent that there is any potential for misunderstanding to begin with.
The next aspect of the unanimity instructions that Coe successfully challenged in the district court is that the jurors were not told that Coe would receive a life sentence if they failed to reach a unanimous sentence. See Tenn. Code. Ann. § 39-2404 (1982) (mandating life sentence if jury is unable to achieve unanimity, but precluding court from informing the jury of this); State v. Simon, 635 S.W.2d 498, 505 (Tenn.), cert. denied, 459 U.S. 1055 (1982). Coe argues that by requiring unanimity for either life or death, a holdout juror preferring a life sentence might vote for death because he thought that a unanimous verdict was required in every case. Therefore, he says, the jury should have been informed of the consequences of a failure by it to achieve unanimity.
We are unpersuaded by this argument. Two circuits have considered and rejected similar arguments regarding similar proceedings and similar state laws. See United States v. Chandler, 996 F.2d 1073, 1089 (11th Cir. 1993); Barfield v. Harris, 540 F. Supp. 451, 472 (E.D.N.C. 1982), aff'd, 719 F.2d 58 (4th Cir. 1983), cert. denied, 467 U.S. 1210 (1984). Coe cites Supreme Court precedent for his proposition, but that precedent is inapt. He cites Beck v. Alabama, 447 U.S. 625 (1980), for the proposition that a jury must be aware of its options if the sentence is to be seen as reliable, but Beck held only that a jury should be informed of its ability to convict of applicable lesser-included offenses, and Coe's jury was so instructed. Beck does not consider the necessity of informing a jury of the consequences of its inability to reach a unanimous verdict.
Such a holding would probably represent a "new rule" for the purposes of Teague v. Lane, 489 U.S. 288 (1989). It is far from clear that current case law requires such a conclusion, much less that it required that conclusion when Coe's conviction and sentence became final in 1984. See Daniels v. Burke, 83 F.3d 760, 764 (6th Cir.) ("Teague constrains our ability to announce new rules in collateral proceedings."), cert. denied, 117 S.Ct. 327 (1996).
There is a fair chance that one of the exceptions to Teague might apply here, namely that "the rule announces a new `watershed' rule of criminal procedure implicating the fundamental fairness and accuracy of the criminal proceeding." In re Green, 144 F.3d 384, 386-87 (6th Cir. 1998). Because we are unsure of Teague's applicability, and because we reject Coe's argument anyway, we will not reach the Teague issue. Because the state has not asserted the Teague issue, we are allowed to so refrain. See Caspari v. Bohlen, 510 U.S. 383, 389 (1994).
Coe also cites Romano v. Oklahoma, 512 U.S. 1, 8-9 (1994), which cited Caldwell v. Mississippi, 472 U.S. 320, 336 (1985), for the proposition that the "jury must not be misled regarding the role it plays in the sentencing decision." True enough. But it does not mislead the jury to impress upon it the importance of unanimity. Romano dealt with the admission of potentially misleading evidence (a previous death sentence imposed on the same defendant), but the Court held that this did not mislead the jury as to its role, or minimize its sense of responsibility. Romano, 512 U.S. at 9.
These necessary indicia are not present in this case either. The jury's role was to deliberate and to attempt to reach a unanimous verdict. The fact that there was a statutorily defined default rule in case the jury could not agree does not change this fact. Indeed, it does not necessarily mislead a jury regarding its role to avoid disclosing what will happen if the jury fails to achieve unanimity. If Coe were correct here, the Supreme Court surely would not have decided Lowenfield v. Phelps, 484 U.S. 231 (1988), the way it did. In that case, the Court approved use of an Allen charge in a capital case. Id. at 237-38. It cited approvingly the Court's statement in Allen that "[t]he very object of the jury system is to secure unanimity by a comparison of views, and by arguments among the jurors themselves." Id. at 237 (quoting Allen v. United States, 164 U.S. 492, 501 (1896)).
The jury was given incomplete information, but not misleading information. Coe's scenario — a minority juror who interprets the unanimity requirement as directing him to give in to the majority — is simply not reasonable. Unanimity means the opposite: that the majority does not win simply because it is a majority. The jury was told precisely this when instructed regarding unanimity at the guilt phase; the jury was instructed to "not surrender your honest conviction . . . for the mere purpose of returning a verdict." The state law did not unconstitutionally deceive the jury and infect the verdict in this case with unreliability, and the district court erred in holding otherwise.
C. Cumulative Effect
Coe and the district court claim that the errors committed by the state trial court cumulatively represent error. Since we have found only one (harmless) error, there is no basis to conclude that there is any cumulative effect here.
Based on the foregoing, then, we REVERSE the district court's grant of habeas corpus relief to Coe.
Coe raises a myriad of issues on cross-appeal. He contends that the district court should have granted, rather than denied, habeas corpus on each of these issues. We affirm the district court, though we rely on different reasons with respect to several of the issues.
Coe petitioned the district court on the grounds that his statement to the police (the heart of the evidence against him) should not have been admitted. He offered many arguments why the statement was suspect: that Coe was both gullible and mentally ill; that the police asked Coe leading questions; that the police knew the details surrounding the crime, raising suspicion that they wrote Coe's statement themselves; that the police continued investigating another suspect even after Coe confessed; that Officer Daniel, who took the statement, allegedly lied to the jury; and that there was evidence that Coe had been physically harmed when he was taken into custody.
The district court rejected those arguments. It applied the standard for voluntariness stated in Miranda v. Arizona, 384 U.S. 436 (1966), and credited various pieces of evidence in the record that supported a finding that the statements were proper. A tape was admitted at trial in which Coe heard his rights and said that he understood them. Later on the tape, he said that he was not mistreated and that he was making the statement of his own free will. Coe signed a Miranda waiver form, which was admitted at trial. The officer who took the statement testified that Coe did not appear to be intoxicated. The district court noted testimony that friends saw Coe with bruises in jail, but rejected that testimony in favor of contrary evidence. Coe says that the discrediting of these witnesses is based on a misunderstanding. He also points to evidence suggesting that police harassed his wife, which allegedly intimidated him into making a statement (even though he was already in custody and there is no evidence that he had any inkling of the alleged harassment). The district court surveyed the evidence, heard from the witnesses first hand, and on the face of the record came to a proper conclusion regarding the alleged coercion underlying Coe's statement.
The district court acted properly in considering coercion rather than credibility. To the extent that Coe argued that his statements were the product of his gullibility and of leading questions, the proper vehicle for such claims is cross-examination at trial. The proper inquiry under Miranda is only whether the government coerced the statement, not whether the statement was the product of Coe's abstract free will. See Colorado v. Connelly, 479 U.S. 157, 169-70 (1986). To the extent that Coe was addressing coercion, the district court's evaluation of the evidence was appropriate.
B. Motion to Amend
Under a scheduling order entered in April 1995, Coe was given until mid-May 1995 to file any amendments. He did so, adding 15 new issues. Discovery was to be completed in July 1995 and dispositive motions filed by August. In December 1995, Coe moved to amend his federal habeas petition a second time, attempting to add 17 new issues. As Coe tells it, this motion was put forward a year before the district court's decision, and seven months before the conclusion of the evidentiary hearing. As the state notes, it was 3 years and 10 months after the filing of the first petition in February 1992.
Seven months later, in July 1996, the district court granted Coe's motion with regard to several law-based claims, but rejected the motion with regard to the other eight fact-based ones. The district court held that the only explanation given by Coe — that he had a new (additional) federal public defender on his team — was insufficient, since he had had ample and expert representation from a federal public defender from the early days of the case, and had had a second lawyer assisting on the case since Summer 1994. The court held that granting the motion fully would unduly prejudice the state, since the state would have to file a large new answer, prepare for a new round of discovery, re-write dispositive motions, and prepare for additional evidentiary hearings. The state claimed that the discovery alone would take three or four months. Reasoning that the legal claims would not spur any new discovery, the district court let them in, while keeping the new fact-based claims out. The court also threw out two legal claims it deemed frivolous: claims that the death penalty in general, and electrocution in particular, violated Coe's rights.
Coe filed a motion for reconsideration, noting that his fact-based claims involved information already known to the state (e.g. Brady claims), so that there would be no discovery burden. He also said that the three to four months of extra discovery projected by the state was both unrealistic and moot, since it took seven months for the district court to resolve the motion; in the meantime the evidentiary hearing had been moved from February to April.
The standard for reviewing denials of motions to amend is abuse of discretion. We have held that:
Under Rule 15(a), leave to amend a pleading shall be freely given when justice so requires. This court has explained the factors that a district court should consider when deciding whether to grant leave to amend. Several elements may be considered in determining whether to permit an amendment. Undue delay in filing, lack of notice to the opposing party, bad faith by the moving party, repeated failure to cure deficiencies by previous amendments, undue prejudice to the opposing party, and futility of amendment are all factors which may affect the decision. Delay by itself is not sufficient reason to deny a motion to amend. Notice and substantial prejudice to the opposing party are critical factors in determining whether an amendment should be granted.
Brooks v. Celeste, 39 F.3d 125, 130 (6th Cir. 1994) (citations omitted and punctuation regularized).
"Delay by itself" is not sufficient to deny a motion to amend, but the delay in this case was substantial. When combined with the prejudice cited by the district court, there were sufficient grounds for the district court to deny the motion. Coe may be right that, in hindsight, there would have been much less prejudice than the state initially predicted if the district court had granted the motion to amend promptly. However, we do not review the district court's hindsight. At the time the district court rendered its decision, allowing amendment of the complaint would have caused prejudice. The district court did not abuse its discretion. Compare Semco, Inc. v. Amcast, Inc., 52 F.3d 108, 114 (6th Cir. 1995) (affirming district court when plaintiff had previously amended its complaint, and moved to amend a mere month before trial after extensive discovery had been conducted) with Security Ins. Co. v. Kevin Tucker Assocs., Inc., 64 F.3d 1001, 1009 (6th Cir. 1995) (reversing district court despite plaintiff's 16-month delay in filing motion to amend, because the action had been dormant for 13 of those months and the case was at its earliest stage of pretrial activity, with discovery cut-off and trial dates not having been set).
C. Ineffective Assistance of Counsel
Coe's next argument is that his trial lawyer: (1) did not sufficiently investigate other suspects, exculpatory evidence, and vital inconsistencies in the testimony of key eyewitnesses; (2) did not investigate or pursue vital physical exculpatory evidence; and (3) did not develop an alibi defense for Coe, which would have been highly plausible.
The specifics of what Coe claims an effective lawyer would have done for him are too voluminous to detail here. They also largely miss the point: just as (or more) important as what the lawyer missed is what he did not miss. That is, we focus on the adequacy or inadequacy of counsel's actual performance, not counsel's (hindsight) potential for improvement. See Sims v. Livesay, 970 F.2d 1575, 1580 (6th Cir. 1992). Furthermore, our review is highly deferential. Ibid.
According to Strickland v. Washington, 466 U.S. 668, 687 (1984), to satisfy the "effectiveness" portion of the test (we need not reach the question of prejudice), Coe would have to demonstrate "that counsel made errors so serious that counsel was not functioning as the `counsel' guaranteed the defendant by the Sixth Amendment." Alternately, combining both elements of the Strickland test, counsel is ineffective if the "trial cannot be relied on as having produced a just result." Id. at 686. In this case, although there are things that Coe's lawyer could have done better, Coe has not convinced us that his lawyer did not function adequately at trial, or that the trial produced an unjust result.
Coe points to exculpatory physical and alibi evidence as examples of things that his lawyer would have pursued were he competent. But, as the district court noted, Coe's lawyer investigated what he could. For instance, the entomological evidence that Coe now marshals (refuting the prosecution's asserted time of death based on the extent of insect infestation on Medlin's corpse) does not seem to have been such an obvious or common part of a defense in the time and place of Coe's trial that counsel was ineffective for neglecting it. Further, while Coe claims that counsel failed to interview numerous alibi witnesses, most of these witnesses were unavailable or would not cooperate with counsel at the time of pre-trial preparation. Others were available, to be sure, but while Coe claims that these witnesses supplied him with an air-tight alibi, we agree with the district court that his lawyer had reasonable doubts about that.
Counsel presented a vigorous insanity defense for Coe which, as a matter of trial strategy, was eminently reasonable. Given the damning confession in the record and the evidence that corroborated it, it was not unconstitutionally ineffective lawyering to put Coe's "wrong man" defense to one side and try to convince the jury that he was responsible but insane. As we have elsewhere noted, the Supreme Court in Strickland "recognized that in the context of ineffective assistance claims based on counsel's failure to investigate, the temptation to rely on hindsight is particularly strong. Consequently, . . . `strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation.'" Sims, 970 F.2d at 1580 (quotation marks omitted) (quoting Strickland, 466 U.S. at 690-91). In this case, the record convinces us that counsel's judgments were similarly reasonable, and that the district court was correct in rejecting Coe's claim.
D. Insufficient Evidence
Coe claims that there was insufficient evidence of premeditation and cool reflection to support his conviction for first-degree common-law murder. Pointing to his own confession, he notes that the record suggests that the murder was a snap reaction to Medlin telling Coe that Jesus loved him. The district court noted, though, that when Coe's initial attempt to kill Medlin by choking her failed, he coolly turned to a different method, the entire deliberate process apparently taking several minutes. In addition, as noted by the district court, Coe appears to have acted with premeditation in selecting his victim. We defer to the district court's reasonable determination of the facts, see 28 U.S.C. § 2254(d)(2), and hold that this evidence suffices.
E. Brady Violations and Prosecutorial Misconduct
Coe's most substantial claim is that the prosecution put on deliberately false testimony, and that it withheld Brady information from him.
1. False testimony
The standard for claims such as this is as follows:
The knowing use of false or perjured testimony constitutes a denial of due process if there is any reasonable likelihood that the false testimony could have affected the judgment of the jury. In order to establish prosecutorial misconduct or denial of due process, the defendants must show (1) the statement was actually false; (2) the statement was material; and (3) the prosecution knew it was false. The burden is on the defendants to show that the testimony was actually perjured, and mere inconsistencies in testimony by government witnesses do not establish knowing use of false testimony.
United States v. Lochmondy, 890 F.2d 817, 822 (6th Cir. 1989) (citations omitted).
Coe's charges center on Officer Alvin Daniel, a key prosecution witness. Coe claims that Daniel knowingly lied about two important pieces of evidence. The first concerns the absence of Coe's tire tracks around the murder scene; Daniel testified that there were no tire tracks from Coe's vehicle because other vehicles' tracks had contaminated the site. However, Coe has not sufficiently proven that Daniel's testimony was false and that the prosecution knew so. At most, the record shows that Daniel's statements fall under the "mere inconsistencies" distinction.
The second item — that there was no physical evidence found in Coe's car — also fails. Daniel implied that there was no evidence found because Coe's vinyl-type seats were the sort you could "wipe off." The problem, as Coe points out, is that the seats were cloth, and appear to stain easily. Once again, though, there is no evidence that Daniel and the prosecution knowingly perpetrated this fabricated testimony (indeed, Daniel himself seemed confused when he testified about it).
Coe's final claim in this category is that the prosecutor lied when he said that his decision to pursue the death penalty was not a quick one. However, statements by counsel are not "testimony." Even if we found this to be false testimony, we would not find that there is a reasonable likelihood that it could have affected the judgment of the jury.
2. Brady violations
The rule of Brady v. Maryland, 373 U.S. 83 (1963), obligates the government "to turn over evidence in its possession that is both favorable to the accused and material to guilt or punishment." Pennsylvania v. Ritchie, 480 U.S. 39, 57 (1987). Materiality is not found by determining (as the district court did in this case) whether there was sufficient evidence to convict even when the exculpatory evidence is added in. Rather, the proper inquiry is whether the Brady violation undermines confidence in the verdict, because there is a reasonable probability that there would have been a different result had the evidence been disclosed. See Kyles v. Whitley, 514 U.S. 419, 434-35 n. 8 (1995).
In this case, Coe says that the state committed a massive Brady violation by not turning over its investigative files to him. The contents include contemporaneous testimony from witnesses who later said different things at trial; inculpatory evidence about other suspects; and exculpatory laboratory results revealing a lack of physical evidence.
The state has a ready and effective reply. First of all, Brady obviously does not apply to information that is not wholly within the control of the prosecution. There is no Brady violation "where a defendant `knew or should have known the essential facts permitting him to take advantage of any exculpatory information,' or where the evidence is available . . . from another source," because in such cases there is really nothing for the government to disclose. United States v. Clark, 928 F.2d 733, 738 (6th Cir.) (quoting United States v. Grossman, 843 F.2d 78, 85 (2d Cir. 1988), cert. denied, 488 U.S. 1040 (1989)), cert. denied, 502 U.S. 885 (1991); see United States v. McMahon, 715 F.2d 498, 501 (11th Cir. 1983), cert. denied, 465 U.S. 1065 (1984). Several of the things about which Coe complains fit into this category, such as the fact that when Coe was searched for scratch marks, he had none, and that a knife taken from his home offered no inculpatory evidence. Another such category is witnesses — people that Coe identified during his statements to the police, and to whom he had as much access as the police. Such information was not under the sole control of the government or improperly kept from Coe.
The next problem for Coe is that it is unclear whether the government did in fact deprive him of these materials. As the state points out, many of the materials that Coe cites appear to have been used by the defense at trial. Coe says, for instance, that he would have subjected certain witnesses to "withering cross-examinations" and thus exposed the inconsistencies in their stories, but the record shows that he did just that, thus suggesting that his lawyer had the materials after all. Indeed, his lawyer testified in the habeas corpus evidentiary hearing that he was confident that the prosecution would have complied with Brady, but that he did not remember what Brady materials he had received, and that his files had been lost. The government has no record of what was handed over to Coe's defense and when. Though it would be a good precautionary policy for the government to keep such records, the burden remains on Coe to prove that the evidence was not disclosed to him. See United States v. Aubin, 87 F.3d 141, 148 (5th Cir. 1996), cert. denied, 117 S.Ct. 965 (1997); Weeks v. Jones, 26 F.3d 1030, 1047 (11th Cir. 1994), cert. denied, 513 U.S. 1193 (1995); United States v. Pedraza, 27 F.3d 1515, 1527 (10th Cir.), cert. denied, 513 U.S. 1004 (1994); Cf. United States v. Driscoll, 970 F.2d 1472, 1482 (6th Cir. 1992) (placing burden on defendant to show that withheld materials contained material exculpatory evidence), cert. denied, 506 U.S. 1083 (1993). We find that he has not met this burden.
Even if Coe had met his burden, we are not convinced that the evidence in question, viewed as a whole, is material. The two most potentially damning sets of allegedly withheld evidence are physical evidence and evidence about a Mr. Gant, the "other" suspect. There were hairs found on the victim's body, for instance, that matched the hair of neither the victim nor Coe. And there was disturbing evidence about Gant — a shifting alibi, a disturbing past, a certain amount of investigation of him, and similar matters — of which Coe could have made good use. Coe's appellate counsel has done an excellent job of presenting this and other evidence to us.
Nevertheless, even if we assume arguendo that this evidence was not disclosed to Coe, and even if we further assume that it should have been, when we view the record as a whole, we cannot conclude that our confidence in the verdict has been undermined. That is, we agree with the district court that there is not a reasonable probability that Coe would have been acquitted had this evidence been disclosed.
Coe would face an uphill battle on this point as well. The information on Gant is only exculpatory if the whole truth about Gant is; if the reason that the police did not turn over inculpatory evidence about Gant is that they had substantial exculpatory evidence to conclude that he was no longer a suspect, Brady would not apply. Brady does not exist to provide fodder for misleading reasonable-doubt defenses.
F. Jury Instructions 1. Instructions on expert testimony, insanity, and parole ineligibility
Coe claims that certain matters of jury instruction — on the treatment of expert testimony, on the disposition of a defendant acquitted by reason of insanity, and on parole ineligibility — were improper. However, these claims are procedurally barred. Coe raised them for the first time in his third state motion for post-conviction relief. The state trial court held that these claims were procedurally barred, because Coe should have raised them earlier. The state trial court's statement suffices as a clear and express statement.
Coe points to the ambiguity of the court of appeals's affirmance, which treated the issues as part of a group that were "waived, previously determined on direct appeal, and/or time barred." Given the trial court's treatment of the claims, however, this does not serve to undo the procedural bar. See Ylst v. Nunnemaker, 501 U.S. 797, 803 (1991) ("[W]here, as here, the last reasoned opinion on the claim explicitly imposes a procedural default, we will presume that a later decision rejecting the claim did not silently disregard that bar and consider the merits.").
Coe has appealed to the state supreme court, but if that court somehow has a basis to reverse this procedural finding (and Coe has given us no basis to conclude that it does), we would hold instead that Coe had not yet exhausted his state remedies. Even in the unlikely event that the supreme court clearly rejects the lower courts' finding of procedural bar (consistent with Ylst), we would still reject Coe's claims, on the merits.
The trial court gave the following instruction on expert testimony:
Expert testimony has been introduced in this case. You should consider this proof in connection with all the other proof in the case, and give it the same consideration as all the other proof, governed by the rules and tests in arriving at the truth. However, expert testimony should be received with caution, while this testimony is sometimes the only means or the best way to arrive at the truth, yet, it is largely a field of speculation beset with many theories and uncertainties, and requires patient and intelligent consideration to determine its truthfulness or falsity.
The instructions for treating general witnesses were not as skeptical, mainly telling the jurors that they were the ultimate decisionmakers, and so should treat the evidence carefully. Another portion of the instructions, on lay witnesses, noted that untrained people are not always good at realizing when someone is mentally diseased or defective, which would seem to bolster the testimony of the experts.
Coe claims that this instruction on expert witnesses "repudiated" his expert testimony, which was an essential part of his insanity defense. He does not explain why he feels that this instruction affected his experts any more negatively than the state's. Given that both sides' evidence on Coe's insanity relied heavily on expert testimony, we will not conclude that this instruction unfairly prejudiced Coe. Coe did have the burden of proof on insanity, to be sure, but we cannot see how directing the jurors toward skepticism and "patient and intelligent consideration" served to "deprive [Coe] of the benefit of [the experts'] testimony." Quercia v. United States, 289 U.S. 466, 471 (1933).
Furthermore, as the district court noted, Tennessee has long used such skeptical instructions. See, e.g., Edwards v. State, 540 S.W.2d 641, 647 (Tenn. 1976), cert. denied, 429 U.S. 1061 (1977). Though it recently softened this language, the Tennessee Court of Appeals noted that it was doing so only prospectively, and noted further that much psychological testimony is still speculative, State v. Givens, No. 01C01-9110-CC-00312, 1993 WL 31710, at *3 (Tenn.Crim.App. Feb. 11, 1993). There is no basis under federal law for attacking these old Tennessee instructions, and so we cannot in this collateral proceeding overrule Tennessee's own law of instructions and evidence.
Regarding the insanity instructions, Coe argues that the jury was given "extraneous information" that biased it in favor of finding him guilty. That is, the jury was allegedly led to believe that if Coe were found not guilty by reason of insanity, he might be able to re-enter society after a very short period of time. The instructions do, indeed, suggest this possibility, though the requisite safeguards (institutionalization and evaluation) are mentioned too.
Coe is correct that the Supreme Court has held, in the context of the federal system, that such instructions should generally not be given, unless, to take one example, a witness or prosecutor stated to the jury that the defendant will "go free" if acquitted by reason of insanity. Shannon v. United States, 512 U.S. 573, 587-88 (1994). The Court also noted in Shannon that the opposite problem may arise, just as Coe has argued — a juror may vote to convict so as to prevent the insane acquitee from going free after the limited hospitalization and evaluation provided for in the statutes. Id. at 586. Nevertheless, the Court counseled against working under the assumption that jurors will not follow their general instruction to base their decisions on the law and facts, and not on extraneous consequences.
For our purposes, the important fact is that the Court's decision in Shannon to limit these instructions came as both a federal prudential matter and a federal statutory one. Id. at 579-84. Because bright-line rules of constitutional law were not at issue (in Shannon or any other case), Coe has not sufficiently demonstrated that the state courts in this case violated a principle of federal law that bound them.
As for parole eligibility, Coe argues that the jury should have been informed that, if he was given a life sentence for the murder, he would not be eligible for parole until he was 113 years old. That is, since his life sentences for rape and kidnapping meant that, under Tennessee law, he would have served 30 years on each sentence before he could be paroled, a life sentence for the murder would have been, effectively, a sentence of life without parole. Knowing this, Coe argues, the jury might have seen a life sentence as a more serious punishment, and therefore sufficient in lieu of a death sentence.
The sole grounding of the district court's rejection of this claim was that the claim was based on a "new rule" under Teague v. Lane, 489 U.S. 288 (1989). We agree that Teague would foreclose this claim if the procedural bar did not, since Coe relies on a case, Simmons v. South Carolina, 512 U.S. 154 (1994), that was decided well after Coe's conviction became final in 1984, and was not sufficiently dictated by prior precedent. See O'Dell v. Netherland, 117 S.Ct. 1969, 1977-79 (1997) (holding that Simmons announced a new rule and retrospective application of the rule was barred by Teague).
Coe's claim fails anyway. There is no reason to assume or conclude that the sentencing judge in Coe's alternative scenario would have ordered Coe to serve his three life sentences consecutively instead of concurrently. See Tenn. R. Crim. P. 32(c)(1) (according such discretion to the trial court). Put another way, a juror would not necessarily have been wrong to believe that a death sentence was the only way to make sure that Coe never left prison alive.
Coe cites Simmons, in which three justices (and three more, concurring in the result) held that a jury should have been told of the defendant's parole ineligibility. There are three important differences between that case and this one, however. First of all, the defendant in Simmons, if convicted, was definitely going to be ineligible for parole. Simmons, 512 U.S. at 156. Second, the prosecution's appeal to the jury for a death sentence was explicitly based on the defendant's future dangerousness, with the implication being that only death would protect the public for sure. Id. at 162. Finally, the defendant in that case specifically requested the instruction at trial. Ibid. Simmons therefore does not apply to this case.
2. Instructions on malice
Coe next contends that his conviction is illegitimate because the jury was not instructed that it must find malice to find murder, even felony murder. Coe bases this argument on the language of the old Tennessee murder statutes, Tenn. Code. Ann. §§ 39-2401-02 (1982), since repealed, which defined murder as requiring malice aforethought and defined felony murder as first-degree murder. He also points to the jury instructions in this case, which said that malice is an element of second-degree murder.
Because we defer to the Tennessee Supreme Court's construction of the elements of Tennessee crimes, Coe's argument is decisively rejected by the clear holding of the Tennessee Supreme Court in State v. Middlebrooks, 840 S.W.2d 317, 335-41 (Tenn. 1992), cert. dismissed, 510 U.S. 124 (1993). In Middlebrooks, a capital case, the Tennessee Supreme Court held that the prosecution does not need to prove malice (or premeditation or specific intent) in order to convict a defendant of felony murder. Id. at 336. The court said that malice is implied from the underlying felony, ibid., but the court did not thereby set up an impermissibly mandatory jury instruction as discussed above. Rather, the court simply explained as a theoretical matter how the required mens rea was automatically supplied in this particular type of murder. Ibid.
G. "Avoiding Arrest or Prosecution" Aggravating Circumstance
One of the aggravating circumstances the jury found was that "the murder was committed for the purpose of avoiding prosecution." Coe argues that there was no evidence, beyond rank speculation, to support this aggravating circumstance. The only direct evidence in the record of Coe's motivations came in his confession, when he said that he snapped when Medlin told him that Jesus loved him. But, as the state supreme court noted, there was testimony at trial (concerning Medlin's personality and Coe's veracity) that cast doubt on the "Jesus" explanation. State v. Coe, 655 S.W.2d 903, 913 (Tenn. 1983), cert. denied, 464 U.S. 1063 (1984). Also, as the prosecutor noted in his closing statement, Coe had a strong motivation to kill Medlin, because she would have been a key witness in Coe's prosecution for kidnapping and rape.
As with other sentencing issues, we review the evidence in the light most favorable to the jury's decision; the finding can be invalidated only if no rational trier of fact could have found the factor beyond a reasonable doubt. Jackson v. Virginia, 443 U.S. 307, 319 (1979). The circumstantial evidence in this case, though thin, is enough to satisfy the standard of review we apply here.
H. Felony-Murder Aggravating Circumstance
Coe challenges the use of the "felony-murder" aggravating circumstance, because the fact that Coe was convicted of killing someone during a felony serves both as a basis for making him a first-degree murderer (and thus eligible for the death penalty), and as an aggravating circumstance that moves him closer to actually receiving the death penalty. Coe cites Middlebrooks for this argument. Middlebrooks held that such double counting is prohibited by the Tennessee Constitution, art. I, § 16, the state's version of the federal Eighth Amendment. Middlebrooks, 840 S.W.2d at 341. The argument accepted by Middlebrooks is that a felony murderer enters sentencing with one strike already against him, even though there is no particular reason why he should receive treatment that is disparate from ordinary common-law premeditated murderers. Id. at 342.
As the district court noted, Middlebrooks is a rule of Tennessee constitutional law, see State v. Howell, 868 S.W.2d 238, 259 n. 7 (Tenn. 1993), cert. denied, 510 U.S. 1215 (1994), and so is not cognizable in federal habeas proceedings. This is true despite Middlebrooks's discussion of federal case law (specifically, Enmund v. Florida, 458 U.S. 782, 797 (1982) (accepting application of death penalty to felony murder only for one who kills, attempts to kill, or intends that lethal force or killing occur); and Tison v. Arizona, 481 U.S. 137, 157-58 (1987) (accepting application of death penalty to felony murder only for one whose personal involvement in the underlying felony is substantial and who exhibits a reckless disregard or indifference to the value of human life)). See Howell, 868 S.W.2d at 259 n. 7 (characterizing Middlebrooks as such).
Indeed, Coe is pursuing his Middlebrooks claim in his third state petition for post-conviction relief. The claim has been rejected by the trial and appeals courts, and is pending before the state supreme court.
However, Coe is not relying wholly on Middlebrooks. Rather, he claims that this "double counting" violates the federal constitution as well. Coe's federal claim rests on the proposition that, "[t]o pass constitutional muster, a capital sentencing scheme must genuinely narrow the class of persons eligible for the death penalty and must reasonably justify the imposition of a more severe sentence on the defendant compared to others found guilty of murder." Lowenfield v. Phelps, 484 U.S. 231, 244 (1988) (quotation marks omitted). There is thus no federalism barrier to raising this claim.
The district court and the state suggest that Coe's claim is foreclosed by Teague v. Lane, 489 U.S. 288 (1989), because he is using Lowenfield to argue for a new rule of law. Coe replies that he is not "breaking new ground" but instead is merely applying the "narrowing" mandate of Zant v. Stephens, 462 U.S. 862, 877 (1983), a case decided six months before Coe's conviction became final. Coe has the better of this argument. He is arguing for nothing more than an application of the plain language of Zant upon which Lowenfield relied ("[A]n aggravating circumstance must genuinely narrow the class of persons eligible for the death penalty and must reasonably justify the imposition of a more severe sentence on the defendant compared to others found guilty of murder.").
On the merits, however, Coe's claim fails.
The district court rejected this claim because it held that Coe was convicted of common-law murder (either alone or in addition to being convicted of felony murder), and so there was no cognizable problem caused by the alleged "double counting." See Carter v. State, 958 S.W.2d 620, 624-25 (Tenn. 1997) (reaching same conclusion on similar set of facts). We are reluctant to affirm on these grounds.
The indictment against Coe charged, in a single count, both common-law murder and felony murder. The jury's verdict was simply one of guilt "as charged in the indictment," and did not distinguish which theory it was convicting under. The jury instructions presented the two theories as alternative bases, and the prosecution told the jury in its closing statement that it could convict under either theory.
Admittedly, it is acceptable for a first-degree murder conviction to be based on two alternative theories even if there is no basis to conclude which one (if only one) the jury used. Schad v. Arizona, 501 U.S. 624, 636-37 (1991); Tenn. Code. Ann. § 40-18-112 (requiring such a result in Tennessee). This does not mean, however, that an alternative conviction such as Coe's represents a conviction on both theories. Put another way, if six jurors believed that Coe was guilty only of felony murder, and six others believed that he was guilty only of common-law murder, it would be proper to say that the jury believed unanimously that Coe was guilty of first-degree murder. It might be improper, by contrast, to say that the jury believed unanimously that Coe was guilty of both forms of murder at the same time.
The state contends that Tennessee law dictates that a conviction like Coe's, where the indictment charges the defendant under both theories, represents a finding of guilt on both theories. Carter, 958 S.W.2d at 625 (holding that such an indictment and conviction cures the Middlebrooks problem). However, there is no evidence of jury instructions or prosecution arguments phrased in the alternative in Carter, and so we cannot conclude that Carter necessarily defeats Coe's "double counting" claim here.
On direct appeal, Coe claimed that he had been subjected to double jeopardy, because he was convicted of both aggravated rape and felony murder predicated on rape. The state supreme court's response to the argument was that there was sufficient evidence in the record to support a conviction for common-law murder, thus preventing any double-jeopardy problem with the first-degree murder conviction. Here too, however, we cannot conclude that the alternative conviction defeats the double-counting claim. If it did, the state could cure double-jeopardy and double-counting problems simply by charging in the alternative, even if there is a basis only for one of the two types of first-degree murder.
This double-jeopardy argument is not before us.
We are not holding that the alternative indictment and conviction cannot cure the double-counting problem. Neither are we holding that Carter cannot do so. Instead, we will address the double-counting issue head on; since we conclude that there is no double-counting problem, there is no need for us to hew a path through the problematic thicket of Carter, Schad, and Tennessee Code § 40-18-112.
Double-counting analysis — the consideration of when a single factor can be used both to make a murderer eligible to receive the death penalty, and as an aggravating circumstance leading to actual imposition of the death penalty — must begin with Lowenfield. In that case, the Supreme Court allowed double counting of the factor of knowingly exposing multiple people to a risk of death. Lowenfield, 484 U.S. at 241-46. There were no other aggravating circumstances found by the jury.
Contrary to Coe's implication on appeal, the Court noted in Lowenfield that narrowing is required at either the eligibility or the imposition stage. Id. at 244-45. That is, a state's system can limit the possible application of the death penalty to certain types of murders, or it can limit the actual imposition of the death penalty to a certain subset of eligible convicts; it must do one of the two, but it need not do both. In allowing double use of the factor, the Court held in Lowenfield that [T]he narrowing function required for a regime of capital punishment may be provided in either of these two ways: The legislature may itself narrow the definition of capital offenses, . . . so that the jury finding of guilt responds to this concern, or the legislature may more broadly define capital offenses and provide for narrowing by jury findings of aggravating circumstances at the penalty phase. Here, the "narrowing function" was performed by the jury at the guilt phase when it found defendant guilty. . . . The fact that the sentencing jury is also required to find the existence of an aggravating circumstance in addition is no part of the constitutionally required narrowing process, and so the fact that the aggravating circumstance duplicated one of the elements of the crime does not make this sentence constitutionally infirm. There is no question but that the Louisiana scheme narrows the class of death-eligible murderers and then at the sentencing phase allows for the consideration of mitigating circumstances and the exercise of discretion. The Constitution requires no more.
Id. at 246 (citation omitted).
The exact same thing can be said of the Tennessee scheme in this case. Admittedly, distinctions can be drawn between the "exposing multiple people to a risk of death" factor in Lowenfield (which clearly narrows the death-eligible pool to more heinous killers) and the felony-murder factor in this case (which arguably sweeps both heinous and relatively non-heinous killers into the pool). But felony murder can function as a proper and permissible narrowing factor at the eligibility stage. In Tison v. Arizona, 481 U.S. 137 (1987), the Supreme Court held that even a non-triggerman convicted of felony murder could constitutionally be executed if he was a major participant in the crime and if he exhibited a reckless disregard for human life. Cf. Enmund v. Florida, 458 U.S. 782 (1982) (reversing death penalty on accomplice to felony murder, and limiting death penalty to defendants who kill, attempt to kill, or at least intend to kill). Thus, if felony murder sometimes has the potential for sweeping less nefarious killers into the pool of the death-eligible, the factor is nevertheless generally permissible. Coe's particular case meets the strictures of both Enmund and Tison. The elements of felony murder for Coe were  killing Medlin  during a rape  when the rape was specifically intended. Therefore, even if Coe's conviction was for felony murder alone, the jury necessarily would have found that all of these elements were met, and so would have found that Coe killed Medlin.
Neither is there anything wrong with using felony murder as an aggravating circumstance. See, e.g., Blystone v. Pennsylvania, 494 U.S. 299, 306-07 (1990). Since, therefore, felony murder can serve to narrow at either the eligibility and imposition stages, and since narrowing need only occur at one of the two stages, there is no double-counting problem here. See Deputy v. Taylor, 19 F.3d 1485, 1497-98, 1500-02 (3d Cir.) (allowing double counting of felony-murder factor), cert. denied, 512 U.S. 1230 (1994); Johnson v. Dugger, 932 F.2d 1360 (11th Cir.) (same), cert. denied, 502 U.S. 961 (1991); Perry v. Lockhart, 871 F.2d 1384, 1393 (8th Cir.) (allowing double-counting of felony-murder factor, and holding that Lowenfield overruled contrary holding in Collins v. Lockhart, 754 F.2d 258 (8th Cir.), cert. denied, 474 U.S. 1013 (1985)), cert. denied, 493 U.S. 959 (1989); Byrne v. Butler, 845 F.2d 501, 515 n. 12 (5th Cir.) (allowing felony murder double-counting because of narrowing guilt-phase element that "offender has specific intent to kill or to inflict great bodily harm and is engaged in the perpetration . . . of . . . armed robbery, or simple robbery."), cert. denied, 487 U.S. 1242 (1988); Julius v. Johnson, 840 F.2d 1533, 1540 (11th Cir.) (allowing double-counting of felony-murder factor), cert. denied, 488 U.S. 960 (1988).
I. Improper Prosecution Arguments
Coe cites four statements made by the prosecution at the sentencing stage that he claims were improper.
First, Coe complains that the prosecutor improperly injected his expertise into his death-penalty argument. The prosecutor said that he took his decision to seek the death penalty very seriously; that the decision was difficult; that he had only ever requested the death penalty once before. Coe cites Eleventh Circuit precedent that is directly on point. See, e.g., Brooks v. Kemp, 762 F.2d 1383, 1410 (11th Cir. 1985) (en banc) (holding very similar comments to be prejudicial). But Kemp ultimately held the error to be harmless, because the jury was well aware of the factors it had to consider, and the proceedings were permeated with similar hand-wringing. Id. at 1414. This case follows that pattern, and we cannot conclude that these comments had a substantial or injurious effect on the jury's verdict. Therefore, even if the comments were improper, they were harmless.
In the subsequent history of Brooks, there was some wrangling between the court of appeals and the Supreme Court over the proper harmless-error standard to be used. See Kemp v. Brooks, 478 U.S. 1016 (1986) (vacating and remanding); Brooks v. Kemp, 809 F.2d 700, 700-01 (11th Cir.) (en banc) (per curiam) (reaffirming previous decision using standard of Chapman v. California, 386 U.S. 18 (1967)), cert. denied, 483 U.S. 1010 (1987). This is irrelevant for our purposes, because we use the Brecht standard (under which it is harder to find that an error was harmful) and so would reach the same result as the Eleventh Circuit in Kemp, a fortiori.
Coe claims next that the prosecutor denigrated mercy, suggesting to the jury that Coe did not deserve any, and allowed the jury to abdicate its responsibility by saying that it represented the community. First, these statements hardly seem so inflammatory that they would have convinced a jury inclined to be merciful to change its mind. Second, when read in context, the prosecutor was not denigrating mercy. Instead, he was responding to arguments by the defense that Coe had had a difficult childhood and so deserved mercy. The prosecutor was simply saying that Coe was responsible for his actions, and that it was for that reason that Coe deserved no mercy. These comments were not improper.
Coe's next argument also fails. The prosecutor said: "[Y]ou're not here as one person, or two persons individually. You're here as representatives of this community, as representatives of justice to do your duty, to be true to your conscience, to be true to God." This was a far cry from saying, as Coe interprets the statement, that the community or God demanded the death sentence.
More problematic is the last prosecutorial statement at issue. In his closing statement, the prosecutor said the following:
I'm not a biblical scholar, ladies and gentlemen, and I don't pretend to be. But I would simply emphasize to you that the whole cornerstone of our law, the law of this land, the law of society is based upon those scriptures where it was established from, and the Bible and the scriptures themselves are replete with those circumstances where capital punishment has been applied. It's applied in reference to both the Old Testament and the New Testament. Whosoever sheddeth man's blood, by man shall his blood be shed. Terms are mentioned regularly throughout the Old and the New. And I just want to ask you to put your mind at rest if that in any way has created any conflict because there's certainly foundation for capital punishment in the Bible and in the scriptures themselves.
Although we find these statements inappropriate, we cannot conclude that they so tainted the proceedings that they constitute reversible error. The cases that Coe cites for the contrary proposition involve cases in which a Bible was in the jury room; there is error in those cases not because the book was the Bible, but because the book was not properly admitted evidence. See Jones v. Kemp, 706 F. Supp. 1534, 1558 (N.D. Ga. 1989); State v. Harrington, 627 S.W.2d 345, 350 (Tenn. 1981), cert. denied, 457 U.S. 1110 (1982).
Therefore, none of these alleged errors represent (either individually or collectively) sufficient grounds to vacate Coe's conviction or sentence.
J. Cumulative Effect
Coe argues that even if none of the errors he alleges are harmful individually, they are harmful in combination with each other. We reject this claim.
K. State Supreme Court Review
Coe claims that he has a federal due-process liberty interest, created by Tennessee in its mandatory death-penalty review statute, Tenn. Code Ann. § 39-2-205 (1982). He claims that by failing to live up to the statutory mandate, the state supreme court contributed to a violation of his due-process rights, and by extension his Eighth Amendment rights.
The statute requires the state supreme court to review a death sentence for (1) arbitrariness; (2) evidence supporting the aggravating and mitigating circumstances found by the jury; and (3) excessiveness or proportionality compared to other death sentences. The supreme court is also empowered by the statute to promulgate such rules as it deems appropriate to carry out the mandate.
Contrary to Coe's claims, however, Tennessee has not created a liberty interest here. To qualify as producing a state-created liberty interest, a statute setting up procedures must put specific limits on official discretion. An implicit requirement recognized by the Supreme Court is including "`explicitly mandatory language,' i.e., specific directives to the decision-maker that if the regulations' substantive predicates are present, a particular outcome must follow, in order to create a liberty interest." Kentucky Dep't of Corrections v. Thompson, 490 U.S. 454, 463 (1989) (promulgating those requirements in the context of prisons). In this case, by contrast, the statute only tells the supreme court what questions it must ask. It does not tell the supreme court how it must do so, and it does not even define the terms (e.g., arbitrariness) of these questions. As a result, Coe has no federal due-process right that was violated by the state supreme court's procedural lapses, if any.
Even if there were a liberty interest here, any error in fulfilling that interest would be harmless. Although the state supreme court did not make detailed findings as to arbitrariness or to the sufficiency of the evidence on the mitigating and aggravating circumstances, our review of the record reveals that the jury's conclusions were neither arbitrary nor lacking in evidentiary support. Furthermore, Coe admits that he has no Eighth Amendment right to proportionality review. Pulley v. Harris, 465 U.S. 37, 44-45 (1984).
L. Exclusion of Women from the Grand Jury
Finally, Coe complains that his Fourteenth Amendment rights were violated because the grand jury that indicted him systematically under-represented women. There is ample evidence in the record that women were very substantially under-represented on the grand jury, apparently because Tennessee used a "key man" system to choose grand jurors. Despite comprising 50.8% of the eligible population, women represented only 29.6% of the grand jury pool (200 of 676 people) from which Coe's grand jury was drawn, a non-random distribution if we have ever seen one (ps.00001). We need not enter into the question of whether there was impermissible exclusion of women, however, because the only question before us is whether Coe had standing to assert this claim. We conclude that he does not, because of Teague.
The district court held that Coe lacked standing. It cited among other cases our decision in Ford v. Seabold, 841 F.2d 677 (6th Cir.), cert. denied, 488 U.S. 928 (1988), which held specifically that a man does not have standing to raise an equal-protection challenge regarding the exclusion of women from a grand jury. The state supreme court had rejected Coe's challenge on the same grounds, five years earlier. State v. Coe, 655 S.W.2d 903, 910 (Tenn. 1983), cert. denied, 464 U.S. 1063 (1984).
Coe responds by noting the Supreme Court's recent decision in Campbell v. Louisiana, 118 S.Ct. 1419 (1998). In Campbell, the Court decided that a white criminal defendant had standing to challenge the exclusion of Blacks from a grand jury, under both equal-protection and due-process theories. Id. at 1424-25.
Regarding the equal-protection claim, the Court cited its prior decision in Powers v. Ohio, 499 U.S. 400, 411 (1991), in which a defendant was found to have third-party standing to raise a Batson challenge to the exclusion from his jury of members of another race. Campbell, 118 S.Ct. at 1422-24. The Court applied Powers, a petit jury case, to the grand jury, and held that "[i]f [the grand jury] process is infected with racial discrimination, doubt is cast over the fairness of all subsequent decisions," which represents injury in fact for Campbell even though he was not a member of the excluded group. Id. at ___ — ___, 118 S.Ct. at 1423-24. The other two parts of third-party standing (common interest, and a first party who is much less likely to vindicate his own interest) were held to apply as well. Id. at 1424.
Unfortunately for Coe, we have already held that the Powers principle of third-party standing represented a new rule, and Coe's conviction became final seven years before Powers was decided. Echlin v. LeCureux, 995 F.2d 1344, 1351 (6th Cir. 1993), cert. denied, 510 U.S. 993 (1994); accord Nguyen v. Reynolds, 131 F.3d 1340, 1351-52 (10th Cir. 1997), cert. denied, 119 S.Ct. 128 (1998); Jones v. Gomez, 66 F.3d 199, 204 (9th Cir. 1995), cert. denied, 517 U.S. 1143 (1996); Van Daalwyk v. United States, 21 F.3d 179, 180 (7th Cir. 1994); Farrell v. Davis, 3 F.3d 370, 372 (11th Cir. 1993) (per curiam). Therefore, the state courts acted reasonably in rejecting Coe's claim of third-party standing. Cf. United States v. Ovalle, 136 F.3d 1092, 1102-04 (6th Cir. 1998) (allowing third-party standing to bring equal-protection claim on direct appeal).
Because Powers represents a new rule, we do not need to determine whether applying Powers and Campbell to gender-based exclusions would itself represent a new rule.
This leaves the due-process portion of Coe's claim. We hold that this too is barred by Teague, though the issue is closer here. Before proceeding, a more detailed account of our task under Teague is in order:
Teague applies when the Supreme Court announces a new rule of criminal procedure. Essentially, if a decision announces a "new rule" of criminal procedure, it is not to be applied retroactively to convictions that have already become final when the decision is announced. . . . A decision announces a new rule if it breaks new ground, imposes new obligations on the states or federal government, or was not dictated by precedent existing at the time the defendant's conviction became final.
In re Green, 144 F.3d 384, 386 (6th Cir. 1998) (citations, quotation marks, and emphasis omitted). One indication that a result is not "dictated by precedent" is when courts have reached divergent results on an issue before resolution by the Supreme Court. See Butler v. McKellar, 494 U.S. 407, 415 (1990).
Green also sets forth the two exceptions to Teague, neither of which apply here:
Two exceptions to the Teague rule, however, permit the retroactive application of a new rule whenever: 1) the rule places certain kinds of primary, private individual conduct beyond the power of the criminal law-making authority to proscribe, or otherwise prohibits imposition of a certain type of punishment for a class of defendants because of their status or offense; or 2) the rule announces a new "watershed" rule of criminal procedure implicating the fundamental fairness and accuracy of the criminal proceeding.
Green, 144 F.3d at 386-87. The first exception clearly does not apply. As for the second exception, Coe has given us no basis to conclude that the gender-balance problem with the grand jury "implicat[es]" the "accuracy" of Coe's indictment.
In Campbell, the Court pointed to two sources in its prior case law for its holding that a defendant can challenge, on due-process grounds, the exclusion of members of another race from a state grand jury. The first was Peters v. Kiff, 407 U.S. 493 (1972). In that case, decided well before Coe's case came along, six justices agreed that the defendant, Peters, was entitled to relief. Three justices (Marshall, Douglas, and Stewart) believed that this entitlement stemmed from the Constitution and from the criminal statute, 18 U.S.C. § 243, that forbids public officials from excluding people from grand jury service because of their race. Peters, 407 U.S. at 497-98. Three other justices (White, Brennan, and Powell) believed that the entitlement stemmed, more narrowly, from the statute alone. Id. at 505-07; see id. at 511 (Burger, J., dissenting) (characterizing concurrence as such); Campbell, 118 S.Ct. at 1424 (same). Justice White wrote in the concurrence that:
By this unambiguous provision, . . . Congress put cases involving exclusions from jury service on grounds of race in a class by themselves. For us the majestic generalities of the Fourteenth Amendment are thus reduced to a concrete statutory command when cases involve race or color which is wanting in every other case of alleged discrimination.
Peters, 407 U.S. at 505-06 (quotation marks omitted) (emphasis added).
Because "the holding of the Court may be viewed as that position taken by those Members who concurred in the judgments on the narrowest grounds," Marks v. United States, 430 U.S. 188, 193 (1977), Peters cannot be said to stand for the proposition that the constitution gave Peters (or Coe) the ability to raise a due-process challenge to the exclusion of Blacks (or women) from his grand jury. Indeed, six justices declined so to hold. Rather, Peters stands only for the proposition that the criminal statute forbidding such exclusion produced the ability to assert such a claim. Unfortunately for Coe, 18 U.S.C. § 243 addressed only race, and there was no parallel gender-based statute in operation at the time of his trial. Therefore, Peters did not establish that Coe was entitled to relief, and a fortiori did not do so as a matter of "existing precedent."
The next source relied upon by Campbell is Hobby v. United States, 468 U.S. 339 (1984). Hobby was decided several months after the Supreme Court denied Coe's writ of certiorari on direct appeal. In Hobby, a white male defendant challenged his indictment because he said that the grand jury excluded Blacks and women. Because Hobby's claim had been dismissed as a matter of law, the Supreme Court assumed that the violation had occurred and proceeded to consider if Hobby had any remedy. Id. at 343. The Court began by noting that purposeful exclusion of women and Blacks from grand jury service was unconstitutional, without distinguishing between gender and race. In proceeding next to the question of remedy, therefore, the Court seemed to be assuming implicitly that Hobby had standing to raise his claim, both on gender and racial grounds, though it noted the narrow holding of Peters. Id. at 343 n. 2. In the end, the Court decided (for reasons that do not concern us) that Hobby was not entitled to a remedy. The Campbell Court read Hobby approvingly, as establishing some sort of due-process protection with regard to race (the only issue Campbell pursued), though it left the determination of the bounds of that protection, which it said were "still open," for the lower court to determine on remand.
We do not doubt that Hobby and Campbell can be read as extending due-process protection to men challenging the exclusion of women, though neither case provided detail on the extent of that protection. The casual manner in which these cases suggest such an extension does not mean, however, that the holdings followed necessarily from "existing precedent." Indeed, the failure of Hobby even to mention the gender/standing question paved the way for conclusions such as the one we reached later in Ford v. Seabold. At any rate, because Hobby was decided after Coe's final appeal to the Supreme Court was turned away, and because it declared a "new rule" if any, Teague bars Coe's claim.
We pause to note additional (though not necessary) evidence that supports this conclusion. First, the Supreme Court denied Coe's petition for certiorari. Ordinarily we would not put much weight on such a denial, but there is more information than usual to be drawn from this particular one. Justices Brennan and Marshall dissented from the denial of Coe's petition, noting that they believed the death penalty to be cruel and unusual punishment. Coe v. Tennessee, 464 U.S. 1063 (1984) (Brennan and Marshall, JJ., dissenting). They did not comment on Coe's prominent objection to the composition of the grand jury that indicted him.
But Coe had not escaped Justice Marshall's notice. He dissented, alone, from the denial of certiorari in another case, Ford v. Kentucky, 469 U.S. 984 (1984), that came after Hobby. In that case, the petitioner asserted a claim identical to Coe's. The Kentucky state courts rejected the gender-based claim on standing grounds, but did not do so on the race-based claim. In significant contrast to the pre-Hobby denial of certiorari in Coe, Justice Marshall addressed head-on the question of standing to bring a due-process claim based on gender, noting the confusion surrounding the issue and citing Coe as an example:
Because the opinion announcing the judgment in Peters was joined by only three Justices, Peters did not definitively resolve the standing question raised in this petition for certiorari. The Court also declined in Alexander v. Louisiana, 405 U.S. 625, 633-34 (1972), to decide whether males could challenge the statutory exemption of women from state grand jury service, although Justice Douglas would have reached the question and invalidated the statute on federal due process grounds. . . . These conflicting pronouncements from the Court and our failure to speak definitively to the issue have spawned the sort of confusion in the lower courts that calls for the exercise of this Court's certiorari jurisdiction. In contrast to the views of the Kentucky Supreme Court, which are shared by the Supreme Court of Tennessee, see State v. Coe, 655 S.W.2d 903 (1983), at least two Federal Courts of Appeals have stated that a male defendant does have a due process right not to have women systematically underrepresented on the state grand jury that indicts him. Gibson v. Zant, 705 F.2d 1543 (11th Cir. 1983); Folston v. Allsbrook, 691 F.2d 184, 186 n. 3 (4th Cir. 1982), cert. denied, 461 U.S. 939 (1983).
Id. at 985-86 (citations omitted and simplified).
There are three interesting things about Justice Marshall's dissent. First, although he is writing in the shadow of Hobby, he does not mention that case, suggesting that it did not serve as a clear statement of Coe's position. Second, Justice Marshall admits that Peters did not resolve the question of standing in the due-process/gender context. Third, Justice Marshall notes that there were lower-court decisions on both sides of the question — evidence of the sort of difference of opinion among reasonable jurists that makes a decision resolving that difference a "new rule."
The difference of opinion among courts was both wider and narrower than Justice Marshall suggested. In addition to the Kentucky and Tennessee Supreme Courts, federal courts, including this one, have held that claimants in Coe's posture lack standing. See, e.g., Ford v. Seabold, 841 F.2d at 687-88 (6th Cir.), cert. denied, 488 U.S. 928 (1988); Beal v. Rose, 532 F. Supp. 306, 311 (M.D. Tenn. 1981), vacated and remanded on other grounds, 703 F.2d 558 (6th Cir. 1982); see also Sheffield v. Lack, 702 F. Supp. 634, 636 (M.D. Tenn. 1987) (citing Beal approvingly); aff'd, No. 88-5459, 1988 WL 121252 (6th Cir. Nov. 15, 1988). See generally United States v. Abell, 552 F. Supp. 316, 320 (D. Me. 1982) (listing cases). However, neither of the circuit-court cases that Justice Marshall cited even mentioned due-process protection; these and many other cases that have extended standing to claimants such as Coe have done so either in the context of federal juries or with regard to equal-protection claims. See generally Abell, 552 F. Supp. at 320 ("conceding that the authority in this area is not unambiguous" and listing cases — of varying applicability — on each side of the question, before finding standing in context of federal equal-protection rights and federal Jury Selection and Service Act).
Thus, when Coe's conviction became final, it was considerably less than clear that he had standing to assert his due-process claim that women were purposely excluded from his grand jury venire. The courts that handled Coe's conviction and direct appeal were not violating any precedent when they determined that Coe did not have standing. Therefore, Teague bars us from applying Campbell retroactively, thereby overruling the determination of the state courts, and from reversing the district court in this case.
Based on the foregoing, we REVERSE the district court insofar as it granted habeas corpus relief, and AFFIRM insofar as it denied relief. Therefore, the award to Coe of habeas corpus relief is REVERSED.
I believe that the jury could reasonably have understood its instructions as requiring it to agree unanimously on the existence of a particular mitigating circumstance before any juror could weigh that circumstance against the aggravating circumstances. I therefore dissent from the majority's reversal of the district court's grant of relief from the death sentence.
The majority correctly states that our inquiry is how a reasonable jury might have understood the instructions. See Mills v. Maryland, 486 U.S. 367, 375-76 (1988). In Mills, the jury was required to indicate which mitigating circumstances it had found by marking "yes" or "no" next to each possible mitigator. The verdict form stated, "[W]e unanimously find that each of the following mitigating circumstances which is marked `yes' has been proven to exist . . . and each mitigating circumstance marked `no' has not been proven." Id. at 387. Although the state claimed that in the event of non-unanimity the jury would leave that space blank, the judge had clearly instructed the jury that, with respect to aggravating circumstances, non-unanimity required a "no" response. See id. at 378. The Supreme Court held that a reasonable jury could have understood the same rule to apply to mitigating circumstances, since the wording of the verdict form was the same in each section. See id. at 378-79. Because the trial judge had done nothing to dispel such a belief, the Court could not "rule out the substantial possibility that the jury may have rested its verdict on the `improper' ground." Id. at 377.
A majority of this court applied Mills in Kordenbrock v. Scroggy, 919 F.2d 1091, 1120-21 (6th Cir. 1990) (en banc) (Kennedy, J., joined by eight others, dissenting from majority opinion that did not reach this issue), cert. denied, 499 U.S. 970 (1991). Kordenbrock characterized the sentencing instructions at issue as requiring unanimity on aggravators but remaining silent as to mitigators. The instructions there stated:
In recommending a sentence . . . you shall consider the following aggravating circumstance, if you believe from the evidence beyond a reasonable doubt that it exists. . . . In recommending a sentence . . . you shall consider such mitigating or extenuating facts and circumstances as have been presented to you in evidence and you believe to be true, including but not limited to such of the following as you believe from the evidence to be true.
Kordenbrock, 919 F.2d at 1108 (Merritt, J., dissenting in part). This court held that "it cannot be reasonably inferred that silence as to finding a mitigating factor would likely cause the jury to assume that unanimity was also a requirement. Indeed it would indicate the opposite." Kordenbrock, 919 F.2d at 1121. In explaining its conclusion, the court distinguished the following instructions, which the Seventh Circuit held improper in Kubat v. Thieret, 867 F.2d 351 (7th Cir.), cert. denied, 493 U.S. 874 (1989):
If, after your deliberations, you unanimously determine that there is no sufficiently mitigating factor or factors to preclude the imposition of the death sentence on the defendant,. . . . If, after your deliberations, you unanimously conclude that there is a sufficiently mitigating factor or factors to preclude the imposition of the death sentence,. . . .
Id. at 1121 (quoting Kubat, 867 F.2d at 369) (first emphasis added by Kordenbrock court; second emphasis added by Kubat court). These instructions were not silent but "specifically told [the jury] that they must find mitigating factors unanimously." Kordenbrock, 919 F.2d at 1121. The Kordenbrock court also noted, with apparent approval, Kubat's holding that these faulty instructions were not cured by other portions of the instructions permitting the jury to state their inability to reach a unanimous verdict. See id. at 1121.
The instructions given at Coe's trial are virtually indistinguishable from those held unconstitutional in Kubat. Coe's jury was told:
If you unanimously determine that . . . [the aggravating circumstances] are not outweighed by any mitigating circumstances, the sentence shall be death. . . . If you unanimously determine that . . . [the aggravating circumstances] are outweighed by one or more mitigating circumstances, the sentence shall be life imprisonment.
Joint Appendix (J.A.) at 1838-39 (emphasis added). Similarly, the verdict form stated, "We, the jury, unanimously find that there are no mitigating circumstances sufficiently substantial to outweigh the [aggravating circumstances]." J.A. at 1839. If one diagrams these sentences, as the state has done, State Reply Br. at 19-20, one can see that they are technically correct, since "unanimously" modifies "determine" (or "find") and refers to the weighing process rather than to the selection of mitigators. However, the same was true in Kubat. Other than using the word "outweighed" instead of "there is no sufficiently mitigating factor," the instructions before us are the same as the Kubat instructions. Both sets of instructions use the word "unanimously" when describing the jury's analysis of mitigating circumstances and are more confusing than mere silence. Three judges of this court recognized this when they expressed their "serious concerns" that these instructions were misleading. Austin v. Bell, 126 F.3d 843, 849 (6th Cir. 1997).
The majority believes it significant that the language quoted above "did not appear in the section of the instructions on finding mitigating factors" but in the later "section on weighing." Ante at 24. The trial judge did not make such a clear division between sections when instructing the jury, and, as Kubat held, the presence of a correct instruction does not necessarily cure an improper instruction. Moreover, taking into consideration the "non-weighing" section of the instructions only increases my certainty that a reasonable jury could have improperly interpreted these instructions. The earlier instructions said:
It is now your duty to determine . . . the penalty. . . . In arriving at this determination, you are authorized to weigh and consider any mitigating circumstances and any of the statutory aggravating circumstances which may have been raised by the evidence. . . .
[N]o death penalty shall be imposed by a Jury but upon an unanimous finding of the existence of one or more of the statutory aggravating circumstances, which shall be limited to the following:. . . .
[I]n arriving at the punishment, the Jury shall consider, as heretofore indicated, any mitigating circumstances which shall include but not be limited to the following:. . . .
J.A. at 1836-38 (emphasis added). While the sentence that introduces the list of mitigators does not contain any form of the word "unanimous," it is not silent about whether unanimity is required. The italicized phrase instructed the jury to consider mitigating circumstances in a manner that had previously been described. This phrase may have been intended to refer generally to the weighing process, which was mentioned in the first quoted sentence. The most natural understanding of this phrase, however, would be that the jury was to apply the parallel instructions it had just received on how to evaluate aggravating circumstances. As in Mills, the parallels between the instructions on aggravators and mitigators would lead the jury to believe that the same approach was required unless something in the instructions signaled a difference. Kordenbrock held that the absence of any reference to unanimity in the instructions on mitigators would be a sufficient signal. Here, however, we have not silence but a specific instruction to incorporate earlier instructions. The most natural earlier instructions to incorporate required unanimity.
The majority notes that the phrase "as heretofore indicated" is quoted from the Tennessee Code. What matters for this case, however, is not what the phrase means in the context of the statute but how it would be understood in the context in which it was presented to the jury.
There is a substantial probability that the jury misunderstood its mandate and thus failed to give proper weight to mitigating circumstances found by some, but not all, of its members. I would therefore affirm the district court's order insofar as it vacated Coe's death sentence on this ground.