Circuit Judge, dissenting:
I would affirm. Agreeing with the majority on all other points, I would hold harmless the error in the charge upon which it reverses this conviction. Rule 52(a), Fed. R.Crim.P.
There was of course technical error in this charge in the respects correctly identified by the majority opinion. I concede that we must be particularly careful in holding harmless any error made in defining to the jury the substantive elements of a crime or a defense when that error has any logical potential for misleading the jury to the defendant’s prejudice. And I have no quarrel with the court’s painstaking mode of analysis. There is no way to assess harmlessness of errors in jury instructions except by assessing their misleading potential in relation to the total context in which they were made and the specific evidence in the case. Ultimately then the estimate of harmlessness or prejudice comes to a judgment call. I accept that the guiding principle for making that judgment call with respect to this non-constitutional error should be whether we believe it “highly probable that the error did not affect the judgment.” R. Traynor, The Riddle of Harmless Error 34-35 (1976). Because I do believe it highly probable that this error did not have that effect here, I simply disagree with the majority on its final judgment call.
Just as the majority concludes that the court’s erroneous, repeated references to defendant’s use of “deadly force” was, in total context, harmless because there was no evidence from which actual use of force rather than mere threat of force could be found, so would I find harmless the trial court’s repeated statements that defendant’s conduct could only be found justified if he reasonably feared “imminent death or serious bodily harm.” The majority concludes that these references foreclosed any finding by the jury that though defendant could not reasonably have feared death or “serious bodily harm” justifying his retaliatory action, he could reasonably have feared mere “bodily injury” that did. Slip op. 318.
While there is a certain logic about this in the abstract, I do not believe it could possibly have operated in this way on the evidence in the case. Defendant’s factual theory of defense was a narrow, simple one— that the correctional officer had actually struck at him with a putty knife and was prepared to do so again. This fact was simply denied by the officer. ..Only if, preliminarily, the jury accepted this fact — or more properly, given the cast of the proof burden, failed to find beyond a reasonable doubt that it did not occur as specifically asserted — could defendant have prevailed *320on the basis of his affirmative defense. Beyond this, under a correct apprehension of the law, the jury must also have accepted— or not found disproved beyond a reasonable doubt — that the retaliatory conduct charged to defendant, including his failure to retreat, was reasonably proportional to the officer’s aggression — found or assumed.
On the evidence in this case it is not only highly improbable, it is inconceivable to me that a jury which accepted — or was in reasonable doubt so that it must assume — the predicate fact of aggression, striking at defendant with a metal knife, would not also have considered the bodily harm threatened by it “serious.” It is therefore inconceivable to me that the jury’s rejection of this defense reflected anything other than a rejection of its narrow factual predicate — on which the evidence was flatly contradictory.
As to the existence of that predicate fact there was of course nothing misleading in the charge. The burden of proof to disprove its occurrence was properly placed upon the state under the proper standard. The charge in its totality — though technically erroneous because of its improvident borrowing from the homicidal self-defense section of the form book — sufficiently conveyed, on the evidence in this case, the critical legal principle: that to constitute justifying self-defense, retaliatory action must (1) be based upon a reasonably feared threat of harm and (2) be reasonably proportional to the threat as perceived. Because the original aggressive act specifically relied upon here so obviously threatened “serious” harm if it threatened any, the technical failure of the court to suggest that though the reasonably perceived threat might not be of “serious” harm it might nevertheless have justified defendant’s retaliatory act here seems to me harmless under the proper test. Obviously so to hold would not condone the giving of such a technically erroneous charge in any but the quite specific situation under review here.
Freed of any concern for a broader precedential effect, or that any general license to use the homicidal self-defense charge in this non-homicidal setting was being given, I would affirm the conviction.