dissenting from IIC,
in which GODBOLD, Chief Judge, KRAVITCH, HATCHETT and CLARK, Circuit Judges, join:The majority holds that the court’s instructions on the elements of malice and intent shifted the burden of proof in violation of Sandstrom v. Montana, 442 U.S. 510, 99 S.Ct. 2450, 61 L.Ed.2d 39 (1979), but finds this error harmless because petitioner raised a non-participation defense rather than arguing that he lacked mens rea. Because the majority’s analysis is based on a false distinction among different types of defenses, and a flawed understanding of the jury’s responsibility with respect to the essential elements of a crime, I dissent.
The plurality opinion of the Court in Connecticut v. Johnson, 460 U.S. 73, 103 S.Ct. 969, 74 L.Ed.2d 823 (1983), cast serious doubt on whether the doctrine of harmless error can be applied to the shifting of a presumption which is so integral to the concept of a fair trial. While this restriction on the harmless error rule failed to gamer the support of a majority, the Court affirmed the pre-existing federal requirement that “before a federal constitutional error can be held harmless, the court must be able to declare a belief that it was harmless beyond a reasonable doubt.” Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 828, 17 L.Ed.2d 705 (1967). This Circuit propounded a similar standard in Lamb v. Jernigan, 683 F.2d 1332, 1342 (11th Cir.1982), cert. denied, 460 U.S. 1024, *1528103 S.Ct. 1276, 75 L.Ed.2d 496 (1983), where it stated that error in a case of a Sandstrom violation can only be harmless if it applied to an element of the crime which was not at issue at the trial, or if the evidence was overwhelming as to the defendant’s guilt.
The majority, however, takes its bearings primarily from Engle v. Koehler, 707 F.2d 241, 246 (6th Cir.1983), aff'd by an equally divided court, — U.S. —, 104 S.Ct. 1673, 80 L.Ed.2d 1 (1984), a Sixth Circuit rendition of the harmless error doctrine which attempts to distinguish between cases in which the defendant claims non-participation in the crime and cases in which the defendant claims lack of mens rea. In Engle the court concluded that, while Sandstrom error in a case where a mens rea defense has been raised is sufficiently prejudicial that error cannot be harmless, similar error in a case in which a non-participation defense has been raised may be harmless. The majority in the instant case justifies this distinction by explaining that where a non-participation defense is presented, intent is not at issue in the trial.
The majority’s analysis fails, at the outset, to recognize that Engle does not compel the result reached in this case. Engle’s, conclusion that a Sandstrom error in a case involving a non-participation defense may be harmless is not determinative of any particular outcome in the instant case. More importantly, neither the Sixth Circuit nor the majority provides an adequate justification for the distinction between non-participation and mens rea defenses. Intent is an issue which must be addressed by the jury in every malice murder case, because it is an element which the jury must find beyond a reasonable doubt if it is to convict a defendant. Intent is thus “at issue,” i.e., remains a question to be resolved by the jury, unless it is admitted by the defendant, see Connecticut v. Johnson, supra, 460 U.S. at 87, 103 S.Ct. at 978, or the evidence concerning it is so “overwhelming” that a jury can reach only one conclusion, see Lamb v. Jernigan, supra, 683 F.2d at 1342. Except where it includes a direct admission of intent, no defense, in and of itself, can take the element of intent out of “issue.” Because the jury must still find this element beyond a reasonable doubt, an instruction which removes the burden of proving this element from the prosecution is no less prejudicial to the defendant in a case where he focuses on the defense of non-participation than in a case where he argues that he lacked mens rea.
The application of the Lamb or Chapman standard to the facts of the case compels rejection of the majority’s conclusion. At no point in the proceeding did petitioner state that he intended to kill the victim. Nor can it be said that the evidence of intent on the part of the petitioner was “overwhelming.”1 Davis stated on several occasions that he had driven the victim and two acquaintances to a private home, that on arrival he had exited to another room, and that he did not even see any of the others until he heard a shot and entered to find the victim on the floor. The State, moreover, presented no alternative version of the facts. The question of whether the prosecution had proved beyond a reasonable doubt petitioner’s intent to kill was very much “at issue.” It is impossible to say that an instruction which relieved the prosecution of its burden with respect *1529to this element was harmless beyond a reasonable doubt.
. The majority proposes to consider instead whether the evidence "that whoever killed the victim did so with intent” was overwhelming. This substitution is wholly unacceptable for the obvious reason that it assumes that “whoever killed the victim” and the petitioner are the same person — the very issue which was before the jury during the trial. Under the "overwhelming evidence" prong of the Lamb test, the court is supposed to evaluate the evidence which was presented at trial, not the conclusions which were drawn from that evidence by the jury. Lamb v. Jernigan, supra, at 1342. The evidence concerning the wounds sustained the victim was not evidence which was offered at trial to describe the mental state of Curfew Davis. It is not appropriate to apply that evidence to him under the Lamb standard simply because the jury later found him guilty. If the retroactive validation and application of all of the State’s evidence to the defendant were acceptable, a reviewing court could find the evidence "overwhelming" in virtually any case in which a guilty verdict had been reached.