Opinion of the Court

1

We can start from some points of agreement. The parties agree that, in Florida, the trial judge is at least a constituent part of "the sentencer" for Clemons purposes, and there is, of course, no doubt that the trial judge "weighed" the coldness factor, as he said in his sentencing order. Nor is there any question that the coldness factor was "invalid" for Clemons purposes, since Parker applied the Clemons rule where a trial judge had weighed two aggravating circumstances that were invalid in the sense that the Supreme Court of Florida had found them to be unsupported by the evidence. See 498 U. S., at 311. It follows that Eighth Amendment error did occur when the trial judge weighed the coldness factor in the instant case. What is in issue is the adequacy of the State Supreme Court's effort to cure the error under the rule announced in Clemons, that a sentence so tainted requires appellate reweighing or review for harmlessness.

2

We noted in Parker that the Supreme Court of Florida will generally not reweigh evidence independently, 498 U. S., at 319 (citing Hudson v. State, 538 So. 2d 829, 831 (per curiam), cert. denied, 493 U. S. 875 (1989); Brown v. Wainwright, 392 So. 2d 1327, 1331-1332 (1981) (per curiam)), and the parties agree that, to this extent at least, our perception of Florida law was correct. The State argues, nonetheless, that, in this case, the State Supreme Court did support the death verdict adequately by performing harmless-error analysis. It relies on the excerpt from the state court's opinion quoted above, and particularly on the second through fourth sentences, as "declar[ing] a belief that" the trial judge's weighing of the coldness factor "was harmless beyond a reasonable doubt" in that it "did not contribute to the [sentence] obtained." Chapman v. California, 386 U. S. 18, 24 (1967). This, however, is far from apparent. Not only does the State Supreme Court's opinion fail so much as to mention

Opinion of the Court

“harmless error," see Yates v. Evatt, 500 U. S. 391, 406 (1991), but the quoted sentences numbered one and five expressly refer to the quite different enquiry whether Sochor's sentence was proportional.

The State tries to counter this deficiency by arguing that the four cases cited following the fourth sentence of the quoted passage were harmless-error cases, citation to which was a shorthand signal that the court had reviewed this record for harmless error as well. But the citations come up short. Only one of the four cases contains language giving an explicit indication that the State Supreme Court had performed harmless-error analysis. See Holton v. State, 573 So. 2d 284, 293 (1990) ("We find the error was harmless beyond a reasonable doubt"). The other three simply do not, and the result is ambiguity.

Although we do not mean here to require a particular formulaic indication by state courts before their review for harmless federal error will pass federal scrutiny, a plain statement that the judgment survives on such an enquiry is clearly preferable to allusions by citation. In any event, when the citations stop as far short of clarity as these do, they cannot even arguably substitute for explicit language signifying that the State Supreme Court reviewed for harmless error.

IV

In sum, Eighth Amendment error occurred when the trial judge weighed the coldness factor. Since the Supreme Court of Florida did not explain or even "declare a belief that” this error "was harmless beyond a reasonable doubt" in that "it did not contribute to the [sentence] obtained," Chapman, supra, at 24, the error cannot be taken as cured by the State Supreme Court's consideration of the case. It follows that Sochor's sentence cannot stand on the existing record of appellate review. We vacate the judgment of the

Opinion of REHNQUIST, C. J.

Supreme Court of Florida and remand the case for proceedings not inconsistent with this opinion.

JUSTICE O'CONNOR, concurring.

It is so ordered.

I join the Court's opinion but write separately to set forth my understanding that the Court does not hold that an appellate court can fulfill its obligations of meaningful review by simply reciting the formula for harmless error. In Chapman v. California, 386 U. S. 18 (1967), we held that before a federal constitutional error can be held harmless, the reviewing court must find "beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained." Id., at 24. This is a justifiably high standard, and while it can be met without uttering the magic words "harmless error," see ante, at 540, the reverse is not true. An appellate court's bald assertion that an error of constitutional dimensions was "harmless" cannot substitute for a principled explanation of how the court reached that conclusion. In Clemons v. Mississippi, 494 U. S. 738 (1990), for example, we did not hesitate to remand a case for "a detailed explanation based on the record" when the lower court failed to undertake an explicit analysis supporting its "cryptic," onesentence conclusion of harmless error. Id., at 753. I agree with the Court that the Florida Supreme Court's discussion of the proportionality of petitioner's sentence is not an acceptable substitute for harmless error analysis, see ante, at 539-540, and I do not understand the Court to say that the mere addition of the words "harmless error" would have sufficed to satisfy the dictates of Clemons.

CHIEF JUSTICE REHNQUIST, with whom JUSTICE WHITE and JUSTICE THOMAS join, concurring in part and dissenting in part.

I join in all that the Court has to say in rejecting Sochor's claim that the application of Florida's "heinousness" factor in

Opinion of REHNQUIST, C. J.

this case violated his constitutional rights. I also agree with the majority that Eighth Amendment error occurred when the trial judge weighed the invalid "coldness" factor in imposing Sochor's death sentence. Accordingly, I join Parts I, II, III-A, and III-B-1 of the Court's opinion. I dissent from Parts III-B-2 and IV of the opinion, however, for I believe that the Supreme Court of Florida cured this sentencing error by finding it harmless. I would thus affirm the judgment below and uphold the sentence.

When a reviewing court invalidates one or more of the aggravating factors upon which the sentencer relied in imposing a death sentence, the court may uphold the sentence by reweighing the remaining evidence or by conducting harmless-error analysis. Clemons v. Mississippi, 494 U. S. 738 (1990). As the majority observes, the Supreme Court of Florida does not in practice independently reweigh aggravating and mitigating evidence, and it did not do so in this case. Ante, at 539-540. In order to sustain Sochor's sentence, the court thus had to find any error harmless. In other words, it had to find beyond a reasonable doubt that the trial judge would still have imposed the death sentence if he had not considered the "coldness" factor when performing the weighing function required by Florida law. Clemons v. Mississippi, supra, at 753; Chapman v. California, 386 U. S. 18, 24 (1967). It seems clear to me that the court reached this conclusion, and that the conclusion is certainly justified by the facts of this case.

After finding that the trial judge erred in relying on the "coldness" factor in determining Sochor's sentence, the Supreme Court of Florida stated:

"The trial court carefully weighed the aggravating factors against the lack of any mitigating factors and concluded that death was warranted. Even after removing the aggravating factor of cold, calculated, and premeditated there still remain three aggravating factors to be weighed against no mitigating circumstances. Striking

Opinion of REHNQUIST, C. J.

one aggravating factor when there are no mitigating circumstances does not necessarily require resentencing. Robinson v. State, 574 So. 2d 108 (Fla. 1991); Holton v. State, 573 So. 2d 284 (Fla. 1990); James v. State, 453 So. 2d 786 (Fla.), cert. denied, 469 U. S. 1098... (1984); Francois v. State, 407 So. 2d 885 (Fla. 1981), cert. denied, 458 U. S. 1122 ... (1982)." 580 So. 2d 595, 604 (1991). The Court now holds that this passage fails to indicate that the error in this case was viewed as harmless. It is true that the passage does not mention the words "harmless error." But we have never held that a court must necessarily recite those words in determining whether an error had an effect on a certain result. In deciding whether the Supreme Court of Florida conducted adequate harmless-error analysis in this case, our focus should not be solely on the particular words and phrases it used to convey its thoughts. Whatever words it used, if they show that it concluded beyond a reasonable doubt that elimination of the "coldness" aggravating factor would have made no difference to Sochor's sentence, then it conducted adequate harmless-error analysis. See Parker v. Dugger, 498 U. S. 308, 319 (1991).

I am convinced by the passage quoted above that the Supreme Court of Florida believed, beyond a reasonable doubt, that the elimination of the "coldness" factor would have made no difference at all in this case. A review of the aggravating and mitigating evidence presented in this case demonstrates why. In making his sentencing determination, the trial judge found four aggravating circumstances, including the "coldness" aggravator. He found absolutely no mitigating evidence. After weighing the four aggravating circumstances against zero mitigating circumstances, the trial judge imposed the death penalty. The Supreme Court of Florida later found the "coldness" aggravating circumstance invalid. It observed, however, that three valid aggravators were left to be balanced against the complete lack of mitigating evidence. On that basis, the court concluded