dissenting..
I agree with the majority that the instruction challenged here had the effect of setting up a conclusive presumption which was not explained away by the totality of the charge. I further agree -with the majority that conclusive presumptions can constitute harmless error. However, I cannot agree with the method of harmless-error analysis employed by the majority and with its conclusion that the . presumption-creating instruction was harmless, Accordingly, I respectfully dissent.
*741 A
In concluding that it is “extremely unlikely that the jury would have relied on [the unconsidered sudden combat] evidence and returned a verdict of manslaughter,” see ante at 740, the majority significantly expands the contours of harmless-error analysis. For, implicit in this facially uncontroversial statement are two radical assumptions: (1) that, in the habeas context, reviewing courts now are obliged to supply missing factual findings; and (2) that, in the habeas context, reviewing courts can and should rely upon evidence that the jury did not consider. In my opinion, neither assumption can be squared with settled authority interpreting the Sixth Amendment jury-trial right and the Due Process Clause.
As the Supreme Court has made clear in a series of recent decisions, an instruction setting up a mandatory presumption engenders an error different in nature than the more typical form of constitutional error — improperly admitted evidence and/or improperly allowed argument. A mandatory presumption directs the jury to presume an element of the crime charged upon finding only certain predicate facts. See, e.g., Sandstrom, 442 U.S. at 517, 99 S.Ct. at 2455. This, of course, directly violates a criminal defendant’s due process rights to have the prosecution prove all elements of the offense charged, see Sullivan v. Louisiana, — U.S. -, -, 113 S.Ct. 2078, 2080, 124 L.Ed.2d 182 (1993) (citing Patterson v. New York, 432 U.S. 197, 210, 97 S.Ct. 2319, 2327, 53 L.Ed.2d 281 (1977) and Leland, v. Oregon, 343 U.S. 790, 795, 72 S.Ct. 1002, 1005, 96 L.Ed. 1302 (1952)), and to have the prosecution persuade the factfin-der beyond a reasonable doubt of the facts necessary to establish each of those elements, id. at -, 113 S.Ct. at 2080-81 (citing In re Winship, 397 U.S. 358, 364, 90 S.Ct. 1068, 1072, 25 L.Ed.2d 368 (1970) and Cool v. United States, 409 U.S. 100, 104, 93 S.Ct. 354, 357, 34 L.Ed.2d 335 (1972) (per curiam)). It also, in my view, tends to undermine the Sixth Amendment jury-trial right. See generally Carella, 491 U.S. at 268-69, 109 S.Ct. at 2422 (Scalia, J., concurring); cf. Sullivan, — U.S. at -, 113 S.Ct. at 2080 (discussing Sixth Amendment right to have the jury, and not the judge, make the requisite finding of guilt).
A conclusive mandatory presumption, as distinguished from a rebuttable mandatory presumption, has a further pernicious effect. By directing, without the possibility of rebuttal, the jury to find the elemental fact merely upon finding certain predicate facts, it “tend[s] to deter a jury from considering any evidence for the presumed fact beyond the predicate evidence.” Yates, 500 U.S. at 406 n. 10, 111 S.Ct. at 1894 n. 10. Indeed, given the “sound presumption of appellate practice[ ] that jurors are reasonable and generally follow the instructions they are given,” id. at 403, 111 S.Ct. at 1893, a reviewing court must assume that the jury did not consider evidence beyond that relating to the predicate facts, because “to do so would be a waste of the jury’s time and contrary to its instructions,” id. at 406 n. 10,1894 n. 10; see also Carella, 491 U.S. at 269, 109 S.Ct. at 2422 (Scalia, J., concurring).
All of this is not to say that a conclusive presumption can never be harmless error. What it does mean, however, as Justice Sca-lia convincingly demonstrates in his concurrence in Carella, is that “the harmless-error analysis applicable in assessing a mandatory conclusive presumption is wholly unlike the typical form of such analysis.” Id. at 267, 109 S.Ct. at 2421. Whereas it makes sense in the case of the more typical form of constitutional error — improperly admitted evidence and/or improperly allowed argument — to perform the type of whole-record “quantitative assessment” outlined in Brecht (and, incidentally, also outlined in Chapman) in order “to determine whether the fact supported by [the] improperly admitted evidence [or improperly allowed argument] was in any event overwhelmingly established by other evidence,” see id., such an inquiry makes no sense where the error is not that the jury may have been swayed by tainted information, but rather is that the jury failed to consider relevant evidence and failed to make a required finding, id. at 267-69,109 S.Ct. at 2421. For, as Justice Scalia explains:
[Such] problem[s] would not be cured by an appellate court’s determination that the record evidence unmistakably established
*742guilt, for that would represent a finding of fact by judges, not by a jury. As with a directed verdict [for the State, which is constitutionally impermissible, see United States v. Martin Linen Supply Co., 430 U.S. 564, 572-73 [97 S.Ct. 1349, 1355, 51 L.Ed.2d 642] (1977) ], “the error in such a case is that the wrong entity judged the defendant guilty.”
Id. at 269, 109 S.Ct. at 2422 (quoting Rose v. Clark, 478 U.S. 570, 578, 106 S.Ct. 3101, 3106, 92 L.Ed.2d 460 (1986)). Thus, the proper question for the reviewing court ‘“is not whether guilt may be spelt out of a record, but whether guilt has been found by a jury according to the procedure and standards appropriate for criminal trials.’ ” Id. (quoting Bollenbach v. United States, 326 U.S. 607, 614, 66 S.Ct. 402, 406, 90 L.Ed. 350 (1946)).
With these principles in mind, Justice Sca-lia has proposed a test for determining whether, despite the presence of a conclusive presumption, a particular case presents a “ ‘rare situation[ ]’ ” where “ ‘[a] reviewing court can be confident that [such a presumption] did not play any role in the jury’s verdict.’ ” Id. 491 U.S. at 270, 109 S.Ct. at 2422 (quoting Connecticut v. Johnson, 460 U.S. 73, 87, 103 S.Ct. 969, 977, 74 L.Ed.2d 823 (1983) (plurality opinion)). Seeking to avoid the specter of factfinding by reviewing courts on the basis of evidence the jury never considered, the Carella test does not direct courts to ascertain whether the presumed fact was otherwise established to varying degrees by the evidence (as the Brecht and Chapman tests would do). Rather, the test instructs reviewing courts to ask (1) whether the instruction established a conclusive pre-. sumption on a charge which did not affect other charges and on which the defendant was acquitted; (2) whether the instruction established a conclusive presumption with respect to an element of the crime which the defendant admitted; or (3) whether
the predicate facts relied upon in the instruction, or other facts necessarily found by the jury, are so closely related to the ultimate fact to be presumed that no rational jury could find those facts without also finding that ultimate fact, making those findings [the] functional[ ] equivalent to ... the element required to be presumed.
Id. at 271, 109 S.Ct. at 2423. If the answer to any of these three questions is “yes,” the error is harmless. See id. Because this test faithfully preserves a criminal defendant’s Sixth and Fourteenth Amendment rights to have an impartial jury make the requisite factual and elemental determinations in his/ her trial, and because it provides assurance that reviewing courts will consider only the evidence that the jury considered, I would apply this test rather than the ill-defined harmless-error test that the majority employs today.
B.
Before applying the Carella test to this case, I feel it appropriate to respond to the arguments against the Carella test and in favor of the whole-record approach outlined in Brecht. Obviously, the most potent of these arguments is the one relied upon by the majority: that the whole-record Brecht analysis applies to “trial - errors,” that the Supreme Court, in a string citation in Fulmi-nante, indicated that a conclusive presumption is “trial error,” and that we therefore are obliged to conduct our harmlessness review in light of the whole record.
I will admit that the Court’s characteriza-: tion of a conclusive presumption as “trial error” in Fulminante is troublesome; indeed, I think this case well illustrates Justice White’s criticism of the “trial error”/“structural error” dichotomy. See Fulminante, 499 U.S. at 291, 111 S.Ct. at 291 (White, J., dissenting in part) (arguing that, in assessing whether harmless-error analysis ought to be applied, courts should disregard the trial error/structural error distinction and instead “consider[] the nature of the right at issue and the effect of [the] error upon the trial”). Despite the aforementioned indication to the contrary, the presence of a non-harmless (as determined by the Carella test) conclusive presumption strikes me as a type of “structural error.” See Carella, 491 U.S. at 268, 109 S.Ct. at 2419 (Scalia, J., concurring) (“The constitutional right to a jury trial embodies a profound judgment about the way in *743which law should be enforced and justice administered. It is a structural guarantee that reflects a fundamental decision about the exercise of official power — a reluctance to entrust plenary powers over the life and liberty of the citizen to one judge or to a group of judges.”) (emphasis supplied) (citations and internal quotation marks omitted); cf. Sullivan, — U.S. at -, 113 S.Ct. at 2083 (denial of the right to a jury verdict of guilt beyond a reasonable doubt is a structural error).
In my view, however, whether we label a conclusive presumption trial error, structural error, or something in between is of no consequence; what matters instead is that we apply the appropriate form of harmless-error review in assessing the effects of that presumption. Applying a whole-record review not only ignores the considerations outlined in Part A. of this dissent, but it also ignores two indications by Supreme Court majorities, subsequent to Fulminante, that the Carella analysis is properly employed by courts reviewing presumptions. See Sullivan, — U.S. at -, 113 S.Ct. at 2082 (indicating, in the direct review context, that Carella provides the proper framework for determining whether or not a mandatory presumption was harmless); Yates, 500 U.S. at 406 n. 10, 111 S.Ct. at 1894 n. 10 (implicitly endorsing, in the habeas context, the Carella test as a means for evaluating whether or not a conclusive presumption was harmless). One might argue that Yates has been superseded by Brecht, and that Brecht did not control in Sullivan because Sullivan was a direct review case. In response, I would point out that Chapman, which did control in Yates and would have controlled in Sullivan had the error therein been deemed amenable to harmless-error review, contemplates a whole-record review every bit as much as Brecht does; nonetheless, the Court has made clear in the Chapman context that, when confronted with presumption error, the typical form of whole-record analysis does not apply. Thus, I read the string citation in Fulmi-nante as merely indicating that a conclusive presumption is amenable to harmless-error review. I do not read it as stating that such a presumption is subject to the usual whole-record harmless-error test ápplieable to most other forms of trial error.
It might also be argued that the Carella test derives from Chapman, see Carella, 491 U.S. at 271, 109 S.Ct. at 2423 (Scalia, J., concurring) (noting that if the Carella test is met, “[t]he error is harmless because it is ‘beyond a reasonable doubt’ that the jury found the facts necessary to support the conviction”) (citing Chapman, 386 U.S. at 24, 87 S.Ct. at 828), and that the replacement of Chapman with Brecht on habeas means that the Carella concurrence has no relevance in habeas cases. In response, I would concede that the Carella test can be theoretically viewed as “deriving from” Chapman. In my view, however, the Carella concurrence can just as easily and fairly be read as (1) explaining. that a conclusive presumption sets up an error which tends to undermine a structural guarantee of the Constitution and which only can be harmless in those “rare” circumstances where the presumption did not play “any role” in the jury’s verdict; (2) setting forth the test for determining whether the error played any such role; and (3) noting, in conclusion and without prior reference to Chapman, that when the Carella test is met, the Chapman test also is met. In light of this, and because abandoning Carella necessarily means that we must welcome factfinding by habeas courts on the basis of evidence the jury did not consider, I prefer the latter reading.
A third argument might be that, in Yates, the Supreme Court has already ratified fact-finding by habeas courts. My response to such an argument simply would be that I agree. As I see it, the. Yates test for reviewing the effects of rebuttable mandatory presumptions, which impermissibly shift the burden of proof from the prosecution to the defendant, does, despite the Supreme Court’s indications to the contrary, reek of factfind-ing by reviewing courts. Cf. generally Yates, 500 U.S. at 411, 111 S.Ct. at 1898 (Scalia, J., concurring in judgment) (explaining that, when a jury has been directed to apply a rebuttable mandatory presumption, it has never found that the prosecution proved the element on which the presumption was erected beyond a reasonable doubt). However, as *744Justice Scalia notes in his Carella concurrence:
It is one thing to say that the effect of th[e] erroneous burden-shifting [effectuated by a rebuttable presumption] will be disregarded if the record developed at trial establishes guilt beyond a reasonable doubt; it is quite another to say that the jury’s failure to make any factual determination of the elemental fact — because of a conclusive presumption resting upon findings that do not establish beyond a reasonable doubt the elemental fact — will be similarly disregarded.
Carella, 491 U.S. at 273, 109 S.Ct. at 2424 (Scalia, J., concurring) (internal quotation marks omitted) (arguing the particular propriety of the Carella test to the conclusive presumption context). Thus, I do not think that the Yates test can and should be read as implicitly endorsing the type of factfinding the majority engages in today.
Finally, one might argue, as does the Brecht majority, that wholesale use of the Brecht test promotes the principles of restraint, often couched in terms of “comity” and “federalism,” underlying the Supreme Court’s more recent habéas jurisprudence. In response, I could only agree if “restraint” is defined solely in terms of state prisoners not being granted very many writs of habeas corpus. For, I think it obvious that factfind-ing on the basis of record evidence that the jury never considered cannot be cited as evidence of judicial restraint. So too do I think it obvious that principles of comity and federalism should never require the continued incarceration of a state prisoner who was not afforded his/her constitutional rights to have an impartial jury make the requisite factual and elemental determinations in his/ her trial just because a federal judge or a panel of, federal judges believe that guilt is “likely” spelt out by the record. Cf. Bollen-bach, 326 U.S. at 614, 66 S.Ct. at 406.
c.
Application of the Carella test to the case at bar easily yields the conclusion that the error here was not harmless.18 I start from the premise that the question of whether an unlawful killing constitutes murder or manslaughter turns on whether or not the killing was committed with malice. See ante note 4. “An intention to inflict injury on the victim which is not justified on any lawful ground or palliated by the existence of any mitigating circumstances is malicious within the meaning of the law.” Commonwealth v. Colon-Cruz, 408 Mass. 533, 562 N.E.2d 797, 808 (1990) (emphasis supplied) (quoting Commonwealth v. McGuirk, 376 Mass. 338, 380 N.E.2d 662, 666-67 (1978), cert. denied, 439 U.S. 1120, 99 S.Ct. 1030, 59 L.Ed.2d 80 (1979)); see also Reddick v. Commonwealth, 381 Mass. 398, 409 N.E.2d 764, 769 (1980) (malice and the presence of legal mitigation are “mutually exclusive”). The presence of “sudden combat” constitutes a mitigating circumstance sufficient to negate malice and to reduce a verdict of murder to manslaughter under Massachusetts law. See Richard, 384 N.E.2d at 638; cf. Commonwealth v. Nardone, 406 Mass. 123, 546 N.E.2d 359, 364 (1989) (distinguishing between assault with intent to murder and assault with intent to kill). And, when such a mitigating circumstance is adequately raised in the evidence (as sudden combat was here), the Commonwealth must prove the absence of this circumstance beyond a reasonable doubt. See Commonwealth v. Nieves, 394 Mass. 355, 476 N.E.2d 179, 182 (1985) (citing Mullaney v. Wilbur, 421 U.S. 684, 697-98, 95 S.Ct. 1881, 1888-89, 44 L.Ed.2d 508 (1975) and Commonwealth v. Stokes, 374 Mass. 583, 374 N.E.2d 87, 94 (1978)).
As the majority opinion states, it is at least reasonably likely that the jurors, on the basis of the challenged instruction, found malice solely upon finding that petitioner stabbed the victim deliberately and cruelly. The foregoing authority, however, makes clear that deliberate and cruel behavior is not nec*745essarily tantamount to malicious behavior. To be specific, the stabbing here could have been both deliberate and cruel, but administered in response to sudden combat, of which there is evidence in this record. Thus, I cannot say that, in this instance, “the predicate facts relied upon in the instruction, or other facts necessarily found by the jury, are so closely related to the ultimate fact to be presumed that no rational jury could find those facts without also finding that ultimate fact.” Carella, 491 U.S. at 271, 109 S.Ct. at 2423. Accordingly, the error had a “substantial and injurious effect or influence in determining the jury’s verdict.” The petition, therefore, should be granted.19
. Obviously, petitioner neither was acquitted of the charge on which the presumption was set up nor admitted at trial that if he did the stabbing, he did so maliciously. Thus, I restrict my inquiry under Carella to whether the predicate facts relied upon in the instruction, or other facts necessarily found by the jury, are so closely related to malice that no rational jury could have found those facts without also finding malice.
. Even were I to employ the deferential standard of review the majority utilizes, I could not join the majority opinion. As I have explained, the error committed here had the effect of deterring the jury from considering, evidence of sudden combat. Yet, there was an abundance of such evidence; indeed, it is undisputed that the stabbing took place in the midst of a drunken melee. In light of this, and in light of the further fact that the Commonwealth bore the burden of proving an absence of sudden combat beyond a reasonable doubt, see Nieves, 476 N.E.2d at 182, I am at a loss to see how the error can be viewed as harmless even under Brecht.