dissenting:
I respectfully dissent.
I agree with the majority that the trial court’s instruction on intent violated Ha-gler’s fourteenth-amendment due process rights by relieving the state of part of its burden of proof in a criminal prosecution, Francis v. Franklin, — U.S. -, 105 S.Ct. 1965, 1977, 85 L.Ed.2d 344 (1985); Sandstrom v. Montana, 442 U.S. 510, 521, 99 S.Ct. 2450, 2458, 61 L.Ed.2d 39 (1979), and that the sole issue for this court to resolve is whether the trial court’s acknowledged “constitutional error in giving the Sandstrom instruction is harmless beyond a reasonable doubt.” 1 The Supreme Court has recently questioned whether giving a Sandstrom instruction can ever constitute harmless error. See Francis v. Franklin, 105 S.Ct. at 1977; Davis v. Kemp, — U.S.-,-, 105 S.Ct. 2689, 2690, 86 L.Ed.2d 707 (1985) (White, J., dissenting from denial of certiorari) (indicating that the Court has had four opportunities recently “to decide whether Sandstrom error may be harmless under any circumstances”); Connecticut v. Johnson, 460 U.S. 73, 103 S.Ct. 969, 978, 74 L.Ed.2d 823 (1983) (plurality opinion) (“[Sandstrom ] error deprived respondent of ‘constitutional rights so basic to a fair trial that their infraction can never be treated as harmless error.’ ”). This circuit has held that Sandstrom error can be harmless under certain limited circumstances. See e.g., In re Hamilton, 721 F.2d 1189, 1190-91 (9th Cir.1983); McGuinn v. Crist, 657 F.2d 1107, 1108-09 (9th Cir.1981), cert. denied, 455 U.S. 990, 102 S.Ct. 1614, 71 L.Ed.2d 850 (1982). , Because I conclude that Hagler’s *717case does not constitute one of those “rare situations in which the reviewing court can be confident that [the] Sandstrom error did not play any role in the jury’s verdict,” Connecticut v. Johnson, 103 S.Ct. at 977 (plurality opinion), I would affirm the district court and grant Hagler’s habeas petition.
This circuit held in In re Hamilton, 721 F.2d 1189 (9th Cir.1983), that unless a defendant clearly concedes or totally fails to dispute the issue of intent at his trial, a “Sandstrom error cannot be held harmless.” Id. at 1192 (emphasis added). This court must therefore “determine whether, by raising a particular defense or by his other actions, [Hagler] himself has taken the issue of intent away from the jury.” Connecticut v. Johnson, 103 S.Ct. at 978 (plurality opinion) (emphasis added). In this case, the magistrate and district judge both concluded, based upon a full review of the record, that “[the] issue of intent was never affirmatively conceded” by Hagler, and that “substantial evidence was introduced [at trial] which could have led a jury to conclude that the homicide, regardless of who did it, was something less than intentional.” I agree with this assessment, and conclude as a result that the trial court’s erroneous Sandstrom instruction was not harmless beyond a reasonable doubt.
The majority’s analysis in this ease is flawed in three respects. First, the majority has overstepped this court’s bounds in reviewing Hagler’s Sandstrom claim by evaluating the quality and quantity of the evidence concerning intent introduced at trial, rather than simply determining whether intent was a disputed issue. The majority correctly notes that under Hamilton, even “a minimum of dispute” over the issue of intent is sufficient “to trigger the corrupting impact of [a] Sandstrom instruction,” and to preclude a determination that giving that instruction was harmless. Yet the majority, although it acknowledges that evidence relating to the issue of intent was introduced at Hagler’s trial, nonetheless concludes that this evidence was “simply insufficient” to overcome the other evidence in the record suggesting that the homicide was intentional. In doing so, the majority violates the principles of both Hamilton and Connecticut v. Johnson, which prohibit courts reviewing Sand-strom claims from examining the issue of intent on the merits once they have concluded that it was disputed:
If the jury may have failed to consider evidence of intent, a reviewing court cannot hold that the error did not contribute to the verdict. The fact that the reviewing court may view the evidence of intent as overwhelming is then simply irrelevant. To allow a reviewing court to perform the jury’s function of evaluating the evidence of intent, when the jury may never have performed that function, would give too much weight to society’s interest in punishing the guilty and too little weight to the method by which decisions of guilt are to be made.
Connecticut v. Johnson, 103 S.Ct. at 977 & n. 15 (plurality opinion) (citations omitted and emphasis added); accord In re Hamilton, 721 F.2d at 1191 (reversal is required if intent is a “disputed issue ... even if substantial evidence of intent exists”) (emphasis added). Thus, once the majority acknowledged that evidence was introduced at trial, no matter how weak, suggesting that the homicide in this case may have been less than intentional, it was required, as a matter of law, to conclude that the trial court’s Sandstrom instruction could not have been harmless beyond a reasonable doubt.
Second, the majority appears to suggest that the fact that Hagler raised an alibi defense at his trial automatically eliminates the issue of intent. However, both the Supreme Court and this court have recognized that an alibi defense and a lack-of-intent defense are not necessarily mutually exclusive. In Connecticut v. Johnson, the Supreme Court stated that “[i]n presenting a defense such as alibi, ... a defendant may in some cases admit that the act alleged by the prosecution was intentional,” thereby making the use of a Sand-strom instruction harmless error. 103 S.Ct. at 978 (plurality opinion) (emphasis *718added). The Court’s statement clearly recognizes that defendants raising alibi defenses do not automatically concede the issue of intent.
Similarly, in McGuinn v. Crist, 657 F.2d 1107 (9th Cir.1981), upon which the majority relies, this court indicated that a defendant’s use of an alibi defense did not automatically prevent him from also raising defenses relating to lack of intent. Although the court ultimately held that the use of a Sandstrom instruction at McGuinn’s trial was harmless error, it expressly distinguished between defenses relating to alibi and those relating to lack of intent, and reached its conclusion only after specifically determining that the defendant failed to contest the issue of intent. Id. at 1108; see In re Hamilton, 721 F.2d at 1191 (concluding that intent was not at issue in McGuinn). This court clearly indicated in McGuinn that the defendant’s claim that he did not perform the shooting was not incompatible with a claim that whoever performed the shooting was not incompatible with a claim that whoever performed the shooting did not intend to kill the victim. At Hagler’s trial, in addition to advancing an alibi defense, the defense introduced substantial evidence relating to the issue of intent. The district court was absolutely correct in perceiving that this immediately distinguishes Hagler’s case from the situation in McGuinn.
Finally, the record belies the majority’s conclusion that intent was not a contested issue at Hagler’s trial.2 Alibi was surely the primary defense advanced by Hagler’s attorney and was obviously the focus of Hagler’s own testimony, but the defense of lack of intent was advanced as well. One must take into account that Hagler’s assertion of an alibi defense prevented him from admitting to shooting the victim and then testifying concerning his reasons for doing so and his state of mind. Nonetheless, the record vividly demonstrates that Hagler’s attorney presented as effective and extensive a lack-of-intent defense as was possible under the circumstances. Regardless of how this court as a factfinder would resolve the intent issue in Hagler’s case, the issue was sufficiently disputed at trial to preclude this court, under the principles of Hamilton and Connecticut v. Johnson, from holding the trial court’s Sandstrom instruction to be harmless error.
Through the testimony of various witnesses, Hagler’s attorney presented evidence from which a jury might conclude that the killing was unintentional. He elicited testimony that the victim, Leif Ellington, regularly drank to excess, and that he was a threatening and dangerous individual when drunk. He introduced Ellington’s autopsy which revealed that Ellington was an alcoholic or at least a chronic drinker, and had Ellington’s acquaintances confirm that he drank frequently and often became “soused,” as one put it. When questioned by Hagler’s attorney, the acquaintances characterized Ellington as “mean,” “mood[y],” and “ornery” when he was drunk, and acknowledged that he could become violent and extremely dangerous in those situations. For example, Alex Man-koski, a defense witness who said that he had been a “friend” of Ellington’s for twenty years, testified that during one such episode about two weeks before Ellington’s death, Ellington had “knocked [his] teeth out and brokefn] his glasses.” Furthermore, Mankoski suggested that this was not an unusual or isolated occurrence.
Other acquaintances of Ellington testified that when he was drunk, he believed “that a lot of people were out to get him____[and] his money.” Still others testified that when he was drunk, Ellington would threaten Lida Pickell, the woman with whom he was living. In fact, a local policeman testified that there had been *719numerous reports to the police of assaults by Ellington on Pickell, and vice versa.
Moreover, Hagler’s attorney introduced testimony concerning Ellington’s great physical strength, as well as his fascination with and occasional abuse of firearms. A local police officer testified that Ellington once “ripped the door off the jail,” and that he had demonstrated great physical strength on several other occasions. The record also reveals that Ellington was over two hundred pounds.
There was also testimony that Ellington had collected a variety of guns, which he kept in his residential trailer, where he was killed. He would occasionally carry one or more of these with him when he went into town, and he was arrested at least once for doing so. An acquaintance of Ellington’s testified that on at least one occasion when he was drunk, Ellington fired his guns off inside his trailer, and the police found a number of bulletholes in the trailer.
Hagler’s attorney introduced evidence that on the day of the murder, Ellington was drinking heavily. By 10:00 A.M., according to Hagler’s testimony, Ellington was beginning on his first pint of whiskey; by about 1:30 P.M., he was beginning on another fifth. By 3:00 or 4:00 P.M., according to Mankoski, Ellington joined him for several glasses of wine. Ellington died sometime during the late afternoon or early evening, and his autopsy revealed a blood-alcohol level of .23 grams percent, indicating that he was “significantly and severely under the influence of alcohol” at the time of death.
The testimony adduced by Hagler’s counsel, at a minimum, raises questions as to whether the killing was intentional. When drunk, Ellington could be “mean,” often turned on his friends and became violent, and occasionally shot off firearms in his trailer; the evidence suggests that Ellington may himself have precipitated the events leading to his death. Because of Ellington’s size, strength, and bullying tendencies, any person with him in his trailer on the afternoon of his shooting could have legitimately feared for his or her safety. Two loaded weapons other than the murder weapon were found in the trailer after Ellington’s death — one, a 32-caliber revolver, was found lying by Ellington’s right hand. Moreover, there may have been a struggle: Ellington had an abrasion above his right eyebrow, with some fresh bleeding from it.
The trajectory of the bullet wounding Ellington in the abdomen was also consistent with the possibility that he was not killed intentionally. In response to questions by Hagler’s attorney, the pathologist who performed Ellington’s autopsy confirmed that Ellington’s abdominal wound could have been sustained as he rose from a sitting position and “c[a]me toward” the shooter. Furthermore, the extremely small dimensions of Ellington’s trailer (12 feet by 6-12 feet by 6V4 feet) could help to explain why Ellington’s wounds were sustained at such point-blank range.
There could have been no other reason for Hagler’s attorney to have adduced the evidence relating to Ellington’s character, his drinking, his physical strength, his assaults on Mankoski and Pickell, his interest in guns and use of them in his trailer, his prior arrest for carrying a weapon, and the trajectory of the bullet wounding him in the abdomen, other than to raise doubts as to whether his murder was intentional. The Washington Rules of Evidence lend added support to the view that intent was in issue. It is likely that much of the character evidence concerning Ellington would have been inadmissible at trial unless it were introduced to support a claim of lack of intent. See Washington Rule of Evidence 404(a)(2) (evidence of character of victim); Fed.R.Evid. 404(a)(2) and Advisory Committee Note on Fed.R.Evid. 404; see also Washington Rules of Evidence 401-402 (relevance). Not one objection to the admission of this evidence was raised by the prosecutor. His silence is eloquent testimony to the fact that intent was in issue.
This court is permitted to examine only whether the issue of intent was disputed, not how well it was disputed or whether there was substantial evidence to support a *720finding of intent. Connecticut v. Johnson, 103 S.Ct. at 977 & n. 15 (plurality opinion); In re Hamilton, 721 F.2d at 1191. The harmless-error analysis of McGuinn, upon which the majority relies, applies only when intent is not disputed by the defendant. See In re Hamilton, 721 F.2d at 1191. In this case, in these circumstances, Hagler disputed the issue of intent as best he could without abandoning his alibi defense. I conclude that the trial court’s unconstitutional Sandstrom instruction was not harmless beyond a reasonable doubt, and I would therefore affirm the district court and grant Hagler’s petition for habeas relief.
. As the majority notes, the Washington Supreme Court concluded that Hagler did not raise lack of intent as a defense at his trial, and implicitly ruled that the trial court's Sandstrom error was harmless beyond a reasonable doubt. See In re Hagler, 97 Wash.2d 818, 650 P.2d 1103, 1109 (en banc). The harmless error determination clearly involved a mixed question of fact and law; would also characterize as a mixed question the issue of whether sufficient evidence was introduced at trial to make lack of intent a disputed issue. See United States v. McConney, 728 F.2d 1195, 1202-03 (9th Cir. 1984). As a result, neither of the Washington Supreme Court’s conclusions involved a strictly "factual issue,” and therefore, neither must be "presumed to be correct" by this court under 28 U.S.C. § 2254(d) (1982) in its habeas review. See Sumner v. Mata, 449 U.S. 539, 544-548, 101 S.Ct. 764, 767-769, 66 L.Ed.2d 722 (1981). Even if the question of whether intent was a disputed issue at trial is treated as a factual determination, and is therefore entitled to section 2254(d)’s "presumption of correctness," I conclude, based upon the record as a whole and especially those portions discussed in this dissent, that the Washington court's determination "is not fairly supported by the record.” 28 U.S.C. § 2254(d)(8); see Sumner v. Mata, 449 U.S. at 547-552, 101 S.Ct. at 769-771.
. Unfortunately, the state trial transcript does not contain a record of the parties’ closing arguments. Such a record might have clarified exactly what defenses and arguments Hagler's attorney advanced at trial, and might well resolve the question of whether the issue of intent was disputed at Hagler’s trial. In the absence of such records, we must extrapolate from the evidence introduced by the defense at trial to determine the theories and arguments it was advancing. See Conway v. Anderson, 698 F.2d 282, 285 (6th Cir.1983).