Page:United States Reports 546.pdf/449

 546US1

238

Unit: $U17

[08-22-08 15:50:12] PAGES PGT: OPIN

BROWN v. SANDERS Breyer, J., dissenting

“In a nonweighing State, so long as the sentencing body ﬁnds at least one valid aggravating factor, the fact that it also ﬁnds an invalid aggravating factor does not in­ fect the formal process of deciding whether death is an appropriate penalty. Assuming a determination by the state appellate court that the invalid factor would not have made a difference to the jury’s determination, there is no constitutional violation resulting from the introduction of the invalid factor in an earlier stage of the proceedings. But when the sentencing body is told to weigh an invalid factor in its decision, a reviewing court may not assume it would have made no difference if the thumb had been removed from death’s side of the scale. When the weighing process itself has been skewed, only constitutional harmless-error analysis or reweighing at the trial or appellate level sufﬁces to guarantee that the defendant received an individualized sentence.” 503 U. S., at 232 (emphasis added). The ﬁrst sentence in this statement is the ﬁrst and only suggestion in our cases that the submission of a vague aggra­ vating circumstance to a jury can never result in constitu­ tional error in a nonweighing State. Indeed, the term “non­ weighing State,” and the signiﬁcance attached to it, does not appear in the Court’s jurisprudence prior to Stringer. The second sentence in the statement is less categorical than the ﬁrst. It suggests that a state appellate court would have to make some form of a harmless-error inquiry to satisfy itself that the invalidated factor “would not have made a difference to the jury’s determination” before it could conclude that there was “no constitutional violation.” Ibid. Given this errant language in Stringer, I agree that it is “[n]ot surpris­ in[g]” that the lower courts have since operated under the assumption “that different rules apply to weighing and non­ weighing States,” and that harmless-error review is neces­ sary only in the former. Ante, at 218, n. 3. My point is simply that such an assumption is unfounded based on our