Chrones v. Pulido - Amicus (Merits)

Docket number: 
No. 07-544
Supreme Court Term: 
2007 Term
Court Level: 
Supreme Court

No. 07-544


In the Supreme Court of the United States






Solicitor General
Counsel of Record
Assistant Attorney General
Deputy Solicitor General
Assistant to the Solicitor
Department of Justice
Washington, D.C. 20530-0001
(202) 514-2217


Petitioner framed the question presented as follows: "Did the Ninth Circuit fail to conform to 'clearly estab lished' Supreme Court law, as required by 28 U.S.C. § 2254(d), when it granted habeas corpus relief by deem ing an erroneous instruction on one of two alternative theories of guilt to be 'structural error' requiring rever sal because the jury might have relied on it?"

The United States will address the underlying ques tion whether a court may apply harmless-error review when a jury is instructed on alternative theories of lia bility, one of which is legally flawed, and the jury re turns a general verdict of guilt.


In the Supreme Court of the United States

No. 07-544











The underlying issue in this case is whether a court may apply harmless-error review when a jury is in structed on alternative theories of liability, one of which is legally flawed, and the jury returns a general guilty verdict. The United States has an interest in that issue because the Court's resolution of it will govern the treat ment of similar instructional errors in federal criminal prosecutions.


Following a jury trial in California state court, re spondent was convicted of first-degree murder, robbery, receiving stolen property, and auto theft. Pet. App. 5a. The jury also found a "special circumstance" of robbery felony-murder. Ibid. He was sentenced to imprison ment for life without parole. Ibid. The state appeals court and state supreme court affirmed the murder con viction. Ibid. The federal district court granted respon dent's petition for a writ of habeas corpus on the claim of prejudicial instructional error on the murder count. Id. at 6a-7a. The court of appeals affirmed. Id. at 1a- 24a.

1. a. On May 24, 1992, sometime between 1 a.m. and 5:30 a.m., Ramon Flores, a gas station cashier, was shot and killed. A neighbor heard a loud bang coming from the direction of the gas station around 3:45 a.m., then a voice yelling as if addressing someone else. The next morning, a cash register taken from the store was found in bushes on the side of a road. Respondent's finger prints were on the cash register, as well as on an un opened can of Coke found on the store counter. The State charged respondent with first-degree murder, robbery, receiving stolen property, and auto theft. Re spondent blamed the robbery and killing successively on various others, including ultimately his uncle Michael Aragon. Pet. App. 2a-3a, 5a.

b. At trial, evidence established that respondent was staying with Aragon and Laura Moore (Aragon's coha bitant). According to Aragon and Moore: Respondent was at home when they went to bed around midnight on May 23, but was gone when they got up at around 3 a.m. to care for their baby. The next morning, respondent showed Aragon his wallet and said, "Look unc, almost all ones." Later that day, Moore discovered respondent was carrying a handgun. At her direction, respondent took the gun apart. Two pieces that Moore had retained to prevent reassembly were later given to police and identified as fitting a .45-caliber Colt (the model match ing the cartridge found at the scene). After seeing a newspaper article about the killing, Aragon asked re spondent if he committed it. Respondent denied he had, but a few days later, when Aragon asked again, respon dent admitted the crime. In a letter from jail, however, respondent wrote to Moore, "If Michael is reading this, tell him I didn't kill that guy, I was just messing with him." Pet. App. 3a-4a.

Respondent testified that Aragon was solely respon sible for the killing. According to respondent: On the night of May 23, after Aragon had smoked cocaine, the two drove to the gas station. Respondent waited outside while Aragon went inside the store ostensibly to pur chase cigarettes. Respondent heard a gunshot and ran into the store. Aragon was holding respondent's gun. Flores was lying on the floor, bleeding from a bullet wound in his face. Respondent yelled at his uncle, ran out of the store, and got in the passenger seat of the car. A few seconds later, Aragon emerged, threw the cash register onto respondent's lap, and drove away. At Aragon's command, respondent pried open the register, gave Aragon the money, and dumped the register. Re spondent denied touching a Coke can in the store that night. Pet. App. 4a-5a.

The prosecutor in closing argued that respondent was guilty of felony murder either because he acted alone as the killer and perpetrator of the robbery, or, on a theory that the prosecutor characterized as a "stretch," because he aided and abetted Aragon, know ing what Aragon intended to do as respondent waited for him outside. RT 1658-1662; Pet. App. 38a. Defense counsel argued that Aragon was the killer and perpetra tor of the robbery and that respondent, who did not know of Aragon's intentions and was coerced into help ing Aragon dispose of the cash register, could be found guilty only as an accessory after the fact with respect to the robbery. RT 1710-1711; Pet. App. 38a.

The court's felony-murder instructions permitted the jury to find respondent guilty of felony murder if he ei ther committed the killing in the course of the robbery or aided and abetted robbery in which a killing occurred. Pet. App. 38a. The felony-murder instructions allowed conviction as an aider and abetter even if the defen dant's involvement in the robbery took place after the killing. Id. at 8a, 51a; see id. at 115a (jury was not in structed that "a felony murder verdict could be based on an aiding and abetting theory only if [respondent] aided and abetted the robbery before the infliction of the fatal wound").1

c. The jury found respondent guilty of first-degree murder, robbery, receiving stolen property, and auto theft. Pet. App. 5a. The jury also returned a "special circumstance" finding under Cal. Penal Code § 190.2(a)(17)(i) (West 1988) that the murder was com mitted while respondent was "ENGAGED IN OR WAS AN ACCOMPLICE IN THE COMMISSION OF OR ATTEMPTED COMMISSION OF ROBBERY." J.A. 56. The jury, however, deadlocked on allegations that respondent personally used a firearm and personally inflicted great bodily injury. Pet. App. 5a. Respondent was sentenced to life imprisonment without parole. Ibid.

2. The California Court of Appeal, 52 Cal. Rptr. 2d 373, and the California Supreme Court, Pet. App. 101a- 120a, affirmed respondent's murder conviction.

The state supreme court agreed with respondent that if one person, acting alone, killed another during the course of a robbery, and a second person thereafter as sisted him in transporting and securing the stolen prop erty, the second person was not guilty of first-degree murder under California law. Pet. App. 101a. Rather, the court held, the killer and accomplice must be jointly engaged in a robbery at the time of the killing. Id. at 102a; see id. at 115a ("[W]e decline to extend our inter pretation of [California's] first degree felony-murder rule to include aiders and abettors or conspirators who join the felonious enterprise only after the murder had been completed.").

The court acknowledged that the jury instructions could have created a contrary "incorrect" implication, explaining:

Unmodified, [Cal. Jury Instructions-Criminal (CALJIC) No. 8.27 (5th ed. 1988)] appears to tell the jury that an aider and abettor in an enumerated fel ony * * * is liable for first degree murder in a kill ing committed by anyone else engaged in the felony. In combination with the * * * instruction concern ing the duration of robbery (CALJIC No. 9.40.1), CALJIC No. 8.27 could well suggest to a jury that a person who aids and abets only in the asportation phase of robbery, after the killing is complete, is nonetheless guilty of first degree murder under the felony-murder rule.

Pet. App. 118a-119a.

The state supreme court did not ultimately decide whether the trial court had a duty to instruct that post- killing aiding of the robbery was insufficient to support a felony-murder conviction, because it found that re spondent could not demonstrate any prejudice from the asserted instructional error. Pet. App. 116a. Specifi cally, the court relied on the "special circumstance" in struction (CALJIC No. 8.80.1 (5th ed. 1993 Supp.)) and verdict (J.A. 56), concluding that the jury "found-ex plicitly, unanimously and necessarily-that defendant's involvement in the robbery, whether as direct perpetra tor or as aider and abettor, commenced before or during the killing of Flores." Ibid.

3. After his state habeas petitions were denied, re spondent filed a petition in federal district court pursu ant to 28 U.S.C. 2254. The district court granted the petition on respondent's claim that the trial court failed to instruct the jury properly that post-killing involve ment does not support felony-murder liability. Pet. App. 38a-67a.

The district court stated that "[i]t was clearly estab lished * * * that errors involving improper instruc tions on a single element of the offense are reversible unless shown to be harmless beyond a reasonable doubt." Pet. App. 44a (citing, e.g., Neder v. United States, 527 U.S. 1, 8-15 (1999)). The court determined, contrary to the California Supreme Court's conclusion, that the error was not harmless under Chapman v. Cali fornia, 386 U.S. 18 (1967), Pet. App. 57a-64a, and con cluded that the error had a "substantial and injurious effect" on the jury's verdict under Brecht v. Abraham son, 507 U.S. 619, 637 (1993), Pet. App. 65a-67a. The district court denied relief on all other claims. Id. at 97a.

4. The Ninth Circuit affirmed in a per curiam deci sion. Pet. App. 1a-24a.

At the outset, the court of appeals noted that the jury deadlocked on allegations that respondent personally used a firearm and personally inflicted great bodily harm; the court accordingly assumed for purposes of its analysis that respondent did not personally murder Flores. Pet. App. 5a. The court stated that the felony- murder instructions erroneously "allowed conviction on the basis of after-the-murder robbery involvement" in contravention of the California Supreme Court's holding that "aiding and abetting a robbery after the killing of a victim does not constitute felony-murder under Califor nia law." Id. at 8a.

The court of appeals agreed with respondent that the state supreme court decision "was contrary to federal law because it improperly applied harmless error analy sis." Pet. App. 11a. The court relied on its decision in Lara v. Ryan, 455 F.3d 1080 (9th Cir. 2006). In Lara, the defendant was convicted of attempted murder based on jury instructions that permitted conviction on either a theory of express malice (a legally proper theory) or implied malice (a legally improper theory). Relying in large part on Sandstrom v Montana, 442 U.S. 510 (1979), the Lara court held that the instructional error was structural and required reversal, unless the review ing court could determine "with absolute certainty" that the defendant was convicted under a proper theory. 455 F.3d at 1086-1087.

The court of appeals found that the jury instructions in this case "le[ft] open the possibility that the jury con victed respondent on a legally impermissible theory, namely, that [respondent] joined the robbery only after Flores was killed." Pet. App. 11a. The court rejected the state supreme court's reliance on the special-circum stances instructions as curative, in light of the admitted error in CALJIC 8.81.17 (5th ed. 1988). The court ob served, as the State conceded, that the instruction "er roneously used the word 'or' rather than 'and' in joining the contemporaneity prong to the 'committed in order to carry out or advance the commission of the crime' prong." Pet. App. 10a. Therefore, according to the court, the instruction "permitted the jury to find the special circumstance that the murder was committed 'while the defendant was engaged in or was an accom plice in' robbery without in fact finding that the acts were contemporaneous." Ibid. The court concluded that, because it could not be "'absolutely certain' that the jury found that [respondent]'s crime of robbery was committed contemporaneously with the murder, the ver dict must be reversed." Id. at 11a-12a.

Judge O'Scannlain filed a special concurring opinion in which he agreed that the case was controlled by Lara, but reasoned that Lara's "attempt to distinguish in structional errors involving impermissible alternative theories from other instructional errors is logically un sustainable and inconsistent with Supreme Court prece dent." Pet. App. 12a-15a. Judge Thomas also filed a concurring opinion, defending Lara's structural-error rule and concluding that respondent would be entitled to relief even under a harmless-error standard. Id. at 15a- 24a.


This case lies at the intersection of two lines of this Court's precedent: Stromberg v. California, 283 U.S. 359 (1931), and successor cases, which hold that if a case is submitted to a jury on alternative theories and one of those theories is legally inadequate, a general verdict of guilt cannot stand; and Rose v. Clark, 478 U.S. 570 (1986), and its successors, which hold that harmless-er ror review applies to omission or misdescription of an offense element in jury instructions. Given the develop ment of this Court's constitutional harmless-error juris prudence and the illogic of applying it to the latter class of cases but not the former, the Court should reconcile its precedents and apply harmlessness review uniformly to both types of instructional errors.

A. The instructional error at issue can easily be de scribed as Stromberg error. Based on the instructions, the jury could have convicted respondent on a valid the ory (that he personally murdered the victim or that he aided and abetted the robbery contemporaneously) or an invalid theory (he aided and abetted the robbery only after the killing occurred). Since post-killing involve ment does not constitute felony murder under California law, it constitutes a defective alternative theory. A jury verdict resting on that theory would violate respondent's constitutional right to a jury finding that the state has proved every element of the crime beyond a reasonable doubt. Accordingly, this case implicates Stromberg's rule of automatic reversal when a general verdict may rest on a constitutionally invalid theory.

B. The Stromberg rule, however, which this Court has deemed contrary to the common law and unworthy of further extension, see Griffin v. United States, 502 U.S. 46, 49-60 (1991), conflicts with later-developed harmless-error jurisprudence. See Chapman v. Califor nia, 386 U.S. 18 (1967). Since Chapman, the Court has repeatedly affirmed that constitutional errors are pre sumptively subject to harmless-error review. Even as suming that the Stromberg line of cases speaks to harm lessness-which is doubtful because it predates Chap man-this Court's post-Chapman decisions applying harmless-error review to omission or misdescription of offense elements undercut the Stromberg rule's continu ing vitality. See, e.g., Rose v. Clark, supra; Neder v. United States, 527 U.S. 1 (1999). The instructional error at issue-the misdescription of the timing element of robbery felony murder-falls squarely within the scope of the Rose/Neder line of cases providing for harmless- error review. It does not rise to the level of the limited class of "structural" errors that necessarily taint the entire trial and render it fundamentally unfair. At the same time, as this case demonstrates, a Rose/Neder er ror can be substantively indistinguishable from a Strom berg error. And it hardly makes sense to apply a less forgiving standard of review when the instruction with the Rose/Neder error is conjoined with a valid alterna tive theory, rather than standing alone. To avoid the "patently illogical" (Pet. App. 13a (O'Scannlain, J., con curring specially)) result that adding a legally valid the ory to a legally invalid theory requires stricter review than if the invalid one stood alone, this Court should apply a single rule of harmless-error review.

C. On direct review, an instructional error should be deemed harmless under Chapman if the reviewing court determines beyond a reasonable doubt that a rational jury would have found the defendant guilty absent the error. Because this case arises on collateral review, the instructional error should be reviewed under the more forgiving standard of Brecht v. Abrahamson, 507 U.S. 619, 630 (1993), under which an error is harmless unless it had "substantial and injurious effect or influence in determining the jury's verdict." Neither Chapman nor Brecht review is limited to what the jury actually found or is dependent on whether the defendant contested guilt under the valid theory. This case should be re manded for application of harmless-error review under Brecht.



This case concerns an error in the jury instructions on one element of a criminal charge-the very type of instructional error to which this Court has consistently applied harmless-error review in a line of cases after Chapman v. California, 386 U.S. 18 (1967). See, e.g., Neder v. United States, 527 U.S. 1, 8-9 (1999); Pope v. Illinois, 481 U.S. 497, 503 (1987); Rose v. Clark, 478 U.S. 570, 581-582 (1986). Specifically, the instructions at is sue failed to explain properly the timing element of the aiding-and-abetting theory of robbery-based felony murder, such that the jury could have found respondent guilty based on either contemporaneous involvement (a valid theory) or post-killing involvement (an invalid the ory). Although the instructional error amounts to an omission or misdescription of an element of the crime- quintessential Rose/Neder error-the Ninth Circuit in stead applied a rule from an older series of cases com mencing with Stromberg v. California, 283 U.S. 359 (1931)-namely, that if a case is submitted to a jury on alternative theories and one of those theories is legally flawed, a general verdict of guilt cannot stand. If the Stromberg rule (and its construction as "structural er ror") were enforced alongside the post-Chapman devel opment of harmless-error doctrine in the Rose/Neder line of cases, it would lead to the anomalous and illogical result that adding a legally valid theory to a legally in valid one makes the error worse. Accordingly, this Court should clarify that harmlessness review uniformly governs all such instructional error.

A. The Instructional Error At Issue Implicates The Stromberg Line Of Cases, Which Is Ripe For Reconsideration

1. The Ninth Circuit relied exclusively on the rea soning of the Stromberg line of cases in concluding that the instructional error was structural and thus required reversal of respondent's murder conviction. In Stromberg, the defendant was charged with one count of violating a California statute that prohibited the public display of a red flag for one of three purposes: opposing government, inviting anarchistic action, or aiding sedi tious propaganda. 283 U.S. at 362-364. The defendant was convicted under a general jury verdict that did not indicate which of the three purposes the defendant had been found guilty of pursuing. After this Court deter mined that it would violate the First Amendment to con vict someone under the first theory of liability (opposing government), it overturned the conviction, holding that "if any of the clauses in question is invalid under the Federal Constitution, the conviction cannot be upheld." Id. at 368. The Court has applied this principle in a se ries of cases (mostly pre-1970) involving general-verdict convictions, where one of the alternative theories upon which the case was submitted was unconstitutional.2

In Yates v. United States, 354 U.S. 298 (1957), over ruled on other grounds by Burks v. United States, 437 U.S. 1 (1978), the Court extended the Stromberg rule to a general verdict in which one of the possible bases of conviction did not independently violate any constitu tional provision, but was otherwise legally flawed. The defendants in Yates were charged with conspiring both to "advocate and teach" the violent overthrow of the United States and to "organize" the Communist Party. Id. at 300. The Court found that the "organizing" object of the conspiracy was legally flawed because the Com munist Party had been "organized" (within the meaning of the statutory prohibition) when it was founded, out side the period set by the statute of limitations. Id. at 311-312. Although the "advocate and teach" object was valid, the Court reversed the conspiracy conviction be cause "the verdict [was] supportable on one ground, but not on another, and it is impossible to tell which ground the jury selected." Id. at 312.

In Griffin v. United States, 502 U.S. 46 (1991), this Court declined to extend its holding in Stromberg and Yates to a general verdict of guilt in a conspiracy charg ing multiple objects where the evidence was insufficient to support guilt as to one of the objects. The Court noted that the rule in Stromberg and Yates was contrary to the common law, under which "a general jury verdict was valid so long as it was legally supportable on one of the submitted grounds-even though that gave no assur ance that a valid ground, rather than an invalid one, was actually the basis for the jury's action." Griffin, 502 U.S. at 49; see id. at 52 (referring to Stromberg as "the fountainhead of decisions departing from the common law"). The Court also criticized Yates as an "unex plained extension" of Stromberg that "explicitly invok[ed] neither the Due Process Clause (which is an unlikely basis) nor our supervisory powers over the pro cedures employed in a federal prosecution." Id. at 55- 56. Stromberg, the Court explained, does "not necessar ily stand for anything more than the principle that, where a provision of the Constitution forbids conviction on a particular ground, the constitutional guarantee is violated by a general verdict that may have rested on that ground." Id. at 53. After noting that "continued adherence to the holding in Yates [was] not at issue in [Griffin]," id. at 56, the Court refused the "unprece dented and extreme" request to extend Yates to situa tions in which the general verdict may have rested on a ground that is factually inadequate (because it is not supported by sufficient evidence) rather than legally inadequate (because, as in Stromberg and Yates, there is some other legal impediment to prosecution), see id. at 56-60. The Court explained that where one possible basis for a jury verdict is factually unsupported, the jury is presumed to have rejected that basis. Id. at 59.

2. Unlike in Griffin, "the continued adherence to the holding[s] in Yates" and Stromberg is potentially at is sue in this case. The instructional error here-though also falling comfortably within the realm of Rose/Neder, see Pt. B(2), infra-can easily be described as a Strom berg/Yates error. The jury was presented with two broad theories under which respondent could be con victed for felony murder: either that he personally killed the victim during commission of the robbery, or that he aided and abetted the commission of the robbery during which the victim was killed. Pet. App. 38a; J.A. 11. The various instructions (see App., infra) suggested that the jury could find respondent guilty under the aiding-and-abetting theory based on either contempora neous involvement in the robbery or post-killing involve ment in the robbery. Pet. App. 11a, 51a, 115a. Because the post-killing involvement is inadequate under Califor nia law to support a conviction for felony murder, the case was submitted on two valid theories (respondent personally killed the victim or aided and abetted the robbery contemporaneously) and one invalid theory (re spondent aided and abetted the robbery only after the killing occurred).

It is true that Stromberg and Yates could be nar rowly distinguished. The legally invalid ground here (non-existent crime) differs from the legally invalid ground in Stromberg (First Amendment-barred crime) and in Yates (time-barred crime). Moreover, unlike with the defective aiding-and-abetting instruction, there was no plausible way to "correct" the defective alternative ground-and thereby apply harmless-error review to that ground-in either Stromberg or Yates.

On the other hand, just as in Stromberg and Yates, the jury's verdict here could have rested on a legally invalid ground. See Evanchyk v. Stewart, 340 F.3d 933, 940 n.2 (9th Cir. 2003) ("[Stromberg and Yates] involved jury instructions for crimes based on facially invalid or legally impossible theories, or 'non-existent' crimes."), cert. denied, 541 U.S. 1067 (2004). And while Yates in volved non-constitutional error, this case (like Strom berg) involves constitutional error. Just as a conviction based on protected speech would violate the First Amendment, conviction for a non-existent crime would violate the Due Process Clause. See, e.g., Fiore v. White, 531 U.S. 225, 228-229 (2001); see also United States v. Gaudin, 515 U.S. 506, 509-510 (1995) (Due pro cess "require[s] criminal convictions to rest upon a jury determination that the defendant is guilty of every ele ment of the crime with which he is charged, beyond a reasonable doubt.") (emphasis added). Moreover, if the legally flawed theory in this case were characterized as felony-murder based on post-killing involvement in the robbery, there is no clear way to "correct" the defective ground. As such, this case implicates the Stromberg rule-which, as discussed next, is wrong on a more fun damental level.

B. This Court's Harmless-Error Jurisprudence, As Applied In The Rose/Neder Line Of Cases, Renders Obsolete Any Rule Of Automatic Reversal For Instructional Error

1. This Court's post-Stromberg/Yates constitutional harmless-error jurisprudence has sharply eroded the precedential value of that line of cases. In Chapman v. California, 386 U.S. 18 (1967), this Court rejected the argument that errors of constitutional dimension neces sarily require reversal of criminal convictions. See id. at 22. Since Chapman, this Court has "repeatedly reaf firmed the principle that an otherwise valid conviction should not be set aside if the reviewing court may confi dently say, on the whole record, that the constitutional error was harmless beyond a reasonable doubt." Dela ware v. Van Arsdall, 475 U.S. 673, 681 (1986); see Ari zona v. Fulminante, 499 U.S. 279, 306-307 (1991) (citing "wide range" of constitutional errors to which this Court has applied harmless-error review). There is a strong presumption that constitutional errors at trial are sub ject to harmless-error inquiry, such that "most constitu tional errors can be harmless." Neder, 527 U.S. at 8 (quoting Fulminante, 499 U.S. at 306); see ibid. ("[I]f the defendant had counsel and was tried by an impartial adjudicator, there is a strong presumption that any other [constitutional] errors that may have occurred are subject to harmless-error analysis.") (quoting Rose, 478 U.S. at 579).

By contrast, only in a "very limited class of cases" will an error be deemed "'structural,' and thus subject to automatic reversal." Neder, 527 U.S. at 8 (quoting Johnson v. United States, 520 U.S. 461, 468 (1997)). Unlike trial errors, structural errors "infect the entire trial process," Brecht v. Abrahamson, 507 U.S. 619, 630 (1993), and "necessarily render a trial fundamentally unfair," Gonzalez v. United States, No. 06-11612 (May 12, 2008), slip op. 11 (quoting Rose, 478 U.S. at 577). See Fulminante, 499 U.S. at 310 ("[A] structural defect affect[s] the framework within which the trial proceeds, rather than simply an error in the trial process itself."). Accordingly, only a select few errors are structural. See United States v. Gonzalez-Lopez, 548 U.S. 140 (2006) (denial of counsel of choice); Sullivan v. Louisiana, 508 U.S. 275, 281-282 (1993) (deficient reasonable-doubt in struction); Gomez v. United States, 490 U.S. 858 (1989) (jury selection by a federal magistrate judge over defen dant's objection); Vasquez v. Hillery, 474 U.S. 254 (1986) (racial discrimination in selection of grand jury); Waller v. Georgia, 467 U.S. 39 (1984) (denial of public trial); McKaskle v. Wiggins, 465 U.S. 168 (1984) (denial of self- representation at trial); Gideon v. Wainwright, 372 U.S. 335 (1963) (complete denial of counsel); Tumey v. Ohio, 273 U.S. 510 (1927) (adjudication by biased judge).

Because Stromberg and Yates preceded Chapman, that line of cases simply does not address the critical question here of whether harmless-error review applies. See Becht v. United States, 403 F.3d 541, 548 (8th Cir. 2005) (Stromberg "establishes that there is 'error'" when "a general verdict may have rested on a ground that is forbidden by the Constitution" but "does not speak to whether the error may be harmless."), cert. denied, 546 U.S. 1177 (2006). As Justice Stewart noted in his opinion concurring in the result in Chapman, be fore that case, this Court had "steadfastly rejected any notion that constitutional violations might be disre garded on the ground that they were 'harmless.'" 386 U.S. at 42-43 (collecting cases). For example, neither the State in Stromberg (see Cal. Br. at 23-27, Stromberg, supra (No. 584)) nor the government in Yates (see Gov't Br. at 94-95, Yates, supra (Nos. 6, 7, 8)) argued that a defendant's conviction may stand, even though the jury's verdict might have rested on a legally flawed ground, if it is clear beyond a reasonable doubt that a rational jury would have found the defendant guilty absent the error.

Accordingly, the Stromberg/Yates line of cases, as the Court observed in Griffin, 502 U.S. at 49-52, should be read only as recognizing an exception to the common- law rule, i.e., that "a general jury verdict was valid so long as it was legally supportable on one of the submit ted grounds-even though that gave no assurance that a valid ground, rather than an invalid one, was actually the basis for the jury's action." Id. at 49. In other words, that line of cases establishes the limited proposi tion that when a general verdict potentially rests on an invalid legal theory, reviewing courts may not presume that the jury relied on a valid alternative theory. Stromberg and its successors do not, however, resolve whether such an instructional error may be subject to harmless-error review. Cf. Sandstrom v. Montana, 442 U.S. 510, 526-527 (1979) (declining to reach issue whether possible Stromberg-type error could be harm less under Chapman).

2. To the extent the Stromberg/Yates line of cases establishes its species of instructional error as struc tural, it cannot be reconciled with post-Chapman devel opments and should be overruled. Instructional errors of the sort at issue here bear no relation to the limited category of pervasive and fundamental errors so intrin sically harmful to the framework of a trial as to be deemed structural, see Pt. B(1), supra, nor do they "vi tiate[] all the jury's findings," Sullivan, 508 U.S. at 281. Accordingly, this Court has repeatedly and consistently held since Chapman that such instructional errors are subject to harmless-error review. See, e.g., Neder, 527 U.S. at 8-9; Pope, 481 U.S. at 503; Rose, 478 U.S. at 582.

In Rose, the trial court instructed the jury in a second-degree murder case that, unless the presumption was rebutted, malice is presumed solely from the fact that "a killing has occurred." 478 U.S. at 574 (citation omitted). Although the Court assumed (id. at 576 n.5) that the instruction violated the defendant's constitu tional right to have his guilt determined beyond a rea sonable doubt by a jury, see Sandstrom, 442 U.S. at 523- 524, the Court held that such an error can be harmless under "Chapman's harmless-error standard." Rose, 478 U.S. at 582. See also, e.g., Yates v. Evatt, 500 U.S. 391, 402 (1991) (unconstitutional mandatory presumption instruction subject to harmless-error review); Carella v. California, 491 U.S. 263 (1989) (per curiam) (same).

The Court has also approved application of the harmless-error doctrine to instructions that erroneously describe elements of the offense, even when conviction based on such instructional error would violate the First Amendment. In Pope, the trial court erroneously in structed the jury that to find the defendant guilty in an obscenity case, it had to find that the material at issue lacked value under "community standards," as opposed to the "reasonable person" standard required by the First Amendment. 481 U.S. at 499-501. Under the in struction, therefore, the jury was permitted to find guilt based on protected speech. This Court nevertheless concluded that the unconstitutional misdescription of the element could be harmless "if a reviewing court con cludes that no rational juror, if properly instructed, could find value in the [defendant's material]." Id. at 503.

The Court has applied the same approach to partial or even complete omission of an element of a criminal offense. In California v. Roy, 519 U.S. 2 (1996) (per curiam), in a case remarkably similar to this one, the state trial court erroneously failed to instruct the jury that it could find the defendant guilty as an aider and abettor only if it found that the defendant had the "in tent or purpose" of aiding the principal's crime. Id. at 3 (emphasis omitted). The Court nonetheless held that "[t]he case before us is a case for application of the 'harmless error' standard." Id. at 5.

Similarly, in Neder, the instructions erroneously failed to include the materiality element of a tax-fraud offense altogether, and thus "prevent[ed] the jury from rendering a 'complete verdict' on every element of the offense." 527 U.S. at 11. The Court nonetheless con cluded that, unlike the constitutional errors it had found to "defy harmless-error review" because they "affect[ed] the framework within which the trial proceeds," id. at 8, an instruction that omits an element of the offense "does not necessarily render a criminal trial fundamentally unfair or an unreliable vehicle for determining guilt or innocence," id. at 9. Applying the same harmless-error inquiry utilized for other trial errors-whether it is "clear beyond a reasonable doubt that a rational jury would have found the defendant guilty absent the error" (id. at 18)-the Court found that the failure to submit the element of materiality to the jury was harmless. Id. at 18-20.

This Court has since reaffirmed Neder and extended harmlessness review to errors in the sentencing context. See Washington v. Recuenco, 126 S. Ct. 2546, 2551-2553 (2006) (holding that Blakely error arising from failure to submit a sentencing factor to the jury is not structural and is subject to harmless-error review); Mitchell v. Esparza, 540 U.S. 12, 15-17 (2003) (per curiam) (uphold ing harmless-error review of State's erroneous failure to charge factual circumstance necessary for imposition of death penalty).

3. As this case amply demonstrates, no logical dis tinction exists between the type of instructional error in the Stromberg/Yates line of cases (suggesting automatic reversal) and that in the more recent post-Chapman line of cases (applying harmless-error review). Conjoining a flawed instruction with a valid one hardly strengthens the case for a rule of automatic reversal. Both types of errors allow a jury to rest its verdict on a legally inade quate theory-that is, a jury might find guilt based on elements that do not constitute a constitutionally per missible crime. Indeed, one could easily characterize the instructional error at issue in this case as falling within the Stromberg/Yates paradigm (allowing convic tion on either valid theory of contemporaneous involve ment or invalid theory of post-killing involvement), as the Ninth Circuit did (Pet. App. 11a), and just as easily characterize it as falling squarely within the scope of the Rose/Neder paradigm (omitting or misdescribing timing element of the involvement), as the district court did (id. at 44a, 51a-52a). The governing rule should not turn on the semantics of how an essentially identical error is characterized. See Neder, 527 U.S. at 10 (refusing to distinguish between instructional error that takes the form of a "misdescription[]" of a broadly drawn element and one that takes the form of an "omission" of a nar rowly drawn one, because "[i]n both cases-misde scriptions and omissions-the erroneous instruction precludes the jury from making a finding on the actual element of the offense").

Accordingly, this Court should reconcile the two lines of cases and clarify-consistent with post-Chapman pre cedents-that harmless-error review applies to all such instructional errors.

4. If any doubt remained as to whether harmless- error should govern the sorts of instructional errors at issue, the illogical implications of the Ninth Circuit's regime should eliminate them. The only noteworthy difference between a Stromberg/Yates error and other instructional error is that, in addition to receiving in structions on an invalid theory of liability, the jury also receives instructions on a valid one. When the instruc tions allow a jury to find guilt based solely on an invalid theory, this Court's harmless-error jurisprudence plainly allows such error to be reviewed for harmless ness. An instructional error on an element does not be come more problematic because the jury may potentially have relied on an alternative theory that was entirely error-free. After all, instructions that present both a valid and an invalid theory of liability reduce the possi bility that an ensuing verdict of guilt rests on an invalid theory. Common sense counsels against a result that permits harmlessness review for the greater, but not the lesser, instructional error. Pet. App. 13a (describing as "patently illogical" the Ninth Circuit's decision that "a jury instruction adding a legally permissible theory to a legally impermissible one somehow increases the grav ity of the error") (O'Scannlain, J., concurring specially).

A comparison of Stromberg and Pope illustrates the incongruity. In Stromberg, the jury might have found guilt on a theory predicated on protected speech (but certainly might not have given the two other valid theo ries). In Pope, the jury necessarily did find guilt on such a theory (given the sole over-inclusive definition of obscenity). Yet, in Pope this Court applied harmless ness review, while in Stromberg it did not. The facts of this case present a similar problem: if the case had been submitted on solely an aiding-and-abetting theory, pre sumably the defect as to the timing or contemporaneity element would have been subject to harmless-error re view under Rose/Neder. It is difficult to conceive any reason why adding a valid direct-perpetrator theory should make the error "structural" and thus subject to automatic reversal.

The Eighth and First Circuits have made that very point when confronted with the problem of reconciling the Stromberg/Yates line of cases with the Rose/Neder line of cases:

[Defendant's assertion] reduces to the strange claim that, because the jury here received both a 'good' charge and a 'bad' charge on the issue, the error was somehow more pernicious than in Rose-where the only charge on the critical issue was a mistaken one. That assertion cannot possibly be right, so it is plainly wrong.

Becht, 403 F.3d at 548 (quoting Quigley v. Vose, 834 F.2d 14, 16 (1st Cir. 1987) (per curiam)).

In Becht, the jury was instructed that child pornog raphy means a "visual depiction [that] is, or appears to be, of a minor engaging in sexually explicit conduct." 403 F.3d at 543 n.2. In light of this Court's decision in Ashcroft v. Free Speech Coalition, 535 U.S. 234, 256 (2002), that portion of the instruction that permitted conviction based on an "appears to be" theory of liability was unconstitutional under the First Amendment. The defendant sought reversal under the Stromberg/Yates line of cases, arguing that the error was structural. In rejecting that claim, the Eighth Circuit held that the error at issue was analogous to the error in cases like Pope: "if Becht's jury had been instructed only on the erroneous theory that Becht possessed images of what 'appeared to be' children, then the error would have been subject to harmless-error review" under this Court's decisions in Pope, Neder, and Rose. Becht, 403 F.3d at 548. The court found that it would therefore be "anomalous to read Stromberg to preclude harmless- error review * * * because the jury also was given the option to convict based on a constitutionally valid theory that Becht possessed images of actual children." Ibid. See Quigley, 834 F.2d at 16 ("Rose plainly trumps Stromberg").

5. In attempting to harmonize the two lines of Su preme Court precedent, two courts of appeals have adopted an odd compromise approach: applying Rose/ Neder harmless-error review to the invalid theory, but precluding reliance on the alternative valid theory as a basis for finding harmlessness. See United States v. Holly, 488 F.3d 1298, 1305-1307 (10th Cir. 2007) ("If the error is harmless as to the erroneously instructed ground considered separately, that ground is no longer insufficient to support the conviction and Stromberg does not require reversal"; but "Stromberg continues to preclude application of harmless error review to the valid ground."); Parker v. Secretary for the Dep't of Corr., 331 F.3d 764, 777-779 (11th Cir. 2003) ("Strom berg cannot foreclose harmless review altogether, be cause an independent basis for a jury verdict is not in sufficient if the relevant error is, considered separately, harmless"; but "[a]n error with regard to one independ ent basis for the jury's verdict cannot be rendered harm less solely because of the availability of the other inde pendent basis."), cert. denied, 540 U.S. 1222 (2004). Un der their approach, the reviewing court may not con sider whether the strength of the evidence on the alter native valid theory is so overwhelming as to render harmless the instructional error on the invalid theory, because any rational jury would have found guilt on the valid theory.3

That approach-while superior to the Ninth Circuit's rule of structural error-falls short. First, it is not clear how harmlessness analysis would operate on an invalid instruction such as post-killing involvement for robbery- felony murder (presumably, the corrected instruction is the same as the alternative valid theory). Second, as suming a reviewing court found the instructional error in the invalid theory not harmless, the court must then ignore the existence of a perfectly valid instruction sub mitted to the jury, even if guilt under that theory were supported by overwhelming evidence. (Suppose, for example, that a defendant were charged with first-de gree murder based on alternative theories of felony- murder (robbery) and intentional murder, see Schad v. Arizona, 501 U.S. 624 (1991), and the felony-murder instruction improperly permitted a finding of guilt based on post-robbery conduct. Even if a videotape revealed that the defendant returned to the scene, drew a gun, and killed the victim in cold blood after the robbery was complete, a court would not be permitted to find the in structional error harmless.) Post-Chapman jurispru dence does not tolerate such a limited inquiry; rather, as this Court stated in Neder, the question is simply whether a rational jury would have returned the same verdict absent the error. See Pt. C.1, infra.

Permitting harmless-error review based solely on the valid alternative ground does not implicate the Court's observation in Neder that a court may not direct a ver dict of guilty. See 527 U.S. at 17 & n.2. A general ver dict that rests on alternative theories, one of which is legally invalid, virtually always results in a jury finding on at least some elements of a valid theory of guilt. (In the felony-murder/intentional murder case described above, for example, the jury must find the element of a killing on any available theory.) Once the jury has re turned a verdict finding at least some elements of the valid theory, finding the error on the alternative theory harmless based on the conclusion that a rational jury necessarily would have found the remaining elements of the valid theory falls squarely within Neder's holding.

* * * * *

In sum, given this Court's unequivocal and consistent holdings since Chapman that harmless-error review governs instructional error when the jury is presented with a single defective theory of liability, precedent and logic dictate that the same rule apply where the jury is instructed on multiple theories of liability and only one of them is defective. A contrary holding would contra dict the last two decades of this Court's harmlessness jurisprudence in the instructional error context.

C. The Instructional Error In This Habeas Case Should Be Reviewed Under The Brecht Standard Of Harmlessness, Where Review Is Not Limited To What The Jury Actu ally Found Or Dependent On Whether The Defendant Contested Guilt Under The Valid Theory

1. Assuming this Court agrees that harmless-error review should govern the sort of instructional error at issue, the question becomes under what standard to con duct the harmlessness review. The Ninth Circuit held that the only instance in which a Stromberg/Yates error can be held harmless is when a reviewing court is "abso lutely certain" that the jury convicted the defendant on the legally valid theory. Pet. App. 11a (citing Lara v. Ryan, 455 F.3d 1080, 1085 (9th Cir. 2006)). Similarly, respondent contends that the instructional error at issue may be deemed harmless only if a reviewing court can ascertain that the jury actually rested its guilty verdict on one of the legally valid theories. Br. in Opp. 11-13. Neder forecloses such a limited view of harmlessness in this context.

In Neder, this Court held that the correct harmless- error inquiry on direct review of instructional error was that articulated in Chapman: "Is it clear beyond a rea sonable doubt that a rational jury would have found the defendant guilty absent the error?" Neder, 527 U.S. at 18. The Court rejected defendant's argument (citing dicta from Sullivan, 508 U.S. at 279) that a finding of harmlessness is limited to situations where the jury ac tually rested its verdict on a valid theory and that a con trary rule would allow judges effectively to direct a guilty verdict. See Neder, 527 U.S. at 17-18. That argu ment, the Court observed, was "simply another form of the argument that a failure to instruct on any element of the crime is not subject to harmless-error analysis." Id. at 17. The Court also explained that such a limited view of harmlessness review "cannot be squared with our harmless-error cases." Id. at 11.4

Nor did the Court in Neder suggest that the avail ability of harmless-error review turns on whether the defendant contested liability under the valid theory. To the contrary, the Court specifically rejected the conten tion that harmlessness is limited to situations "where the defendant admitted the element on which the jury was improperly instructed." Neder, 527 U.S. at 13. The Court concluded on the facts of that case, "where the omitted element was uncontested and supported by overwhelming evidence," that the erroneous instruction was harmless. Id. at 17-19. The Court, however, did not hold conversely that omission of a contested element can never be harmless. See United States v. Neder, 197 F.3d 1122, 1129 & n.6 (11th Cir. 1999) (on remand) ("[T]he Supreme Court [in Neder] did not hold that omission of an element can never be harmless error un less uncontested."), cert. denied, 530 U.S. 1261 (2000). Otherwise, harmless-error review could be defeated at will by a defendant's ipse dixit. The Court made that point plain in Neder when it stated: "If, at the end of [a thorough examination of the record], the court cannot conclude beyond a reasonable doubt that the jury ver dict would have been the same absent the error-for example, where the defendant contested the omitted element and raised evidence sufficient to support a con trary finding-it should not find the error harmless." 527 U.S. at 19 (emphasis added). See Rose, 478 U.S. at 583-584 ("[O]ur harmless-error cases do not turn on whether the defendant conceded the factual issue on which the error bore. Rather, we have held that 'Chap man mandates consideration of the entire record.' * * * Thus, the fact that [defendant] denied that he had [the requisite intent] does not dispose of the harmless- error question.") (quoting United States v. Hasting, 461 U.S. 499, 509 n.7 (1983)).

2. On habeas review, courts apply the more forgiving standard set forth in Kotteakos v. United States, 328 U.S. 750, 776 (1946): whether the error "had substantial and injurious effect or influence in determining the jury's verdict." Brecht, 507 U.S. at 623; see, e.g., Roy, 519 U.S. at 5-6 (holding that Brecht, not Chapman, gov erned claim of instructional error raised on federal ha beas review). Under that standard, a jury verdict should be reversed only if a court has "grave doubt" about whether the verdict would have been the same absent the error. O'Neal v. McAninch, 513 U.S. 432, 436-437 (1995). As under the stricter Chapman stan dard, of course, review is not limited to what the jury actually found or dependent on whether the defendant contested guilt under the valid theory.

Just last term, the Court confirmed that the Brecht standard applies in Section 2254 cases, even where (like here) the state court did not itself apply the Chapman standard. See Fry v. Pliler, 127 S. Ct. 2321, 2325 (2007). In Fry, this Court held that the habeas court need not apply both Chapman and Brecht; because the more stringent Chapman standard is necessarily subsumed within the relaxed Brecht standard, the habeas court need only apply the latter. See id. at 2327 ("[I]t cer tainly makes no sense to require formal application of both tests (AEDPA/Chapman and Brecht) when the lat ter obviously subsumes the former."). After Fry, there can be no doubt that non-structural errors-including the instructional error at issue-are subject on collat eral review to the Brecht standard, such that the error is harmless as long as it did not have "a substantial and injurious effect" on the jury's verdict.

3. The United States does not take any position on whether the instructional error at issue was harmless under Brecht on the facts of this particular case. Rather, like petitioner (Br. 38), the government sug gests that the Court follow its normal practice of re manding the case for application of the Brecht standard in the first instance. See, e.g., Dobbs v. Zant, 506 U.S. 357, 359 n.* (1993) (per curiam) ("[W]e see no reason to depart here from our normal practice of allowing courts more familiar with a case to conduct their own harmless- error analyses.").5


The judgment of the Ninth Circuit should be re versed.

Respectfully submitted.



Solicitor General
Assistant Attorney General
Deputy Solicitor General
Assistant to the Solicitor


1 Relevant excerpts from the jury instructions appear in the appen dix to this brief.

2 See, e.g., Leary v. United States, 395 U.S. 6, 31-36 (1969) (conviction for transporting marijuana may have rested on alternate theory as to which the jury received an unconstitutional presumption instruction); Terminello v. Chicago, 337 U.S. 1, 5 (1949) (disorderly conduct convic tion may have rested on alternate ground of constitutionally protected speech); Cramer v. United States, 325 U.S. 1, 29 (1945) (two of three overt acts submitted as basis for treason conviction did not satisfy the constitutional requirement that a treason conviction be based "on the Testimony of two Witnesses to the same overt Act"); Williams v. North Carolina, 317 U.S. 287 (1942) (bigamy conviction may have rested on alternate theory that North Carolina did not have to recognize defen dant's Nevada divorce decree in violation of the Constitution's Full Faith and Credit Clause).

Stromberg has appeared only sporadically in post-1970 decisions. See, e.g., Zant v. Stephens, 462 U.S. 862, 880-881 (1983) (restating Stromberg's rule in the capital context, but holding that the jury's verdict finding multiple aggravating factors was valid despite the legal invalidity of one of them). In Mills v. Maryland, 486 U.S. 367 (1988), the Court cited Stromberg, but the case did not involve multiple theories of liability; rather, the jury was instructed on a single theory as to the relevance of mitigating circumstances in capital sentencing. See id. at 375. This Court reversed the death sentence because it was unclear whether the instruction was interpreted correctly (permitting individual juror consideration of any mitigating circumstance) or incorrectly (requiring unanimity before consideration of a particular mitigating circumstance)-and not because the jury was presented with alternate valid and invalid theories. In any event, as the this Court made clear, the capital-sentencing context drove the reversal. See id. at 376 ("In reviewing death sentences, the Court has demanded even greater certainty that the jury's conclusions rested on proper grounds.").

3 The Tenth Circuit does recognize two exceptions to its rule: where the jury has explicitly or necessarily made findings that support convic tion on the valid ground, or where there is no evidence to support conviction on the invalid ground. In either scenario, the court must pre sume that the jury relied on the valid ground and affirm the conviction. See Holly, 488 F.3d at 1306 n.5. The government agrees with the results produced by both exceptions, but not the general rule.

4 As noted above (p. 27, supra), a Stromberg/Yates error does not ordinarily provide any greater concern of an impermissible directed verdict than Rose/Neder error. This case illustrates the point. If a court were to find harmless the instructional error in this case (permitting conviction based only on post-killing involvement) because compelling evidence existed to satisfy the contemporaneity element of aiding-and-abetting robbery felony-murder, that finding would no more constitute a directed verdict than if the timing element had been omitted in its entirety, as in Neder. The jury still found a killing and the defendant's culpable mental state, i.e., intent to commit robbery.

5 Although the district court did apply Brecht (after applying Chapman) in reviewing for harmlessness, Pet. App. 65a-67a, the court of appeals did not do so, id. at 11a-12a. Judge Thomas's concurring opinion conducted harmless-error review, but only under the stricter Chapman standard. Id. at 16a-23a.



Court of Appeal No. A065850
Superior Court No. SC29805




Volume 2



Filed: May 2, 1994


* * * * *





(Penal Code s. 189)


If a human being is killed by any one of several per sons engaged in the commission or attempted commis sion of the crime of robbery, all persons, who either di rectly and actively commit the act constituting such crime, or who with knowledge of the unlawful purpose of the perpetrator of the crime and with the intent or pur pose of committing, encouraging, or facilitating the com mission of the offense, aid, promote, encourage, or insti gate by act or advice its commission, are guilty of mur der in the first degree, whether the killing is intentional, unintentional, or accidental.

[J.A. 11]



CALJIC 8.80.1 (1990 New)
(Penal Code s. 190.2)


[J.A. 12]

* * * * *


If you find that a defendant was not the actual killer of a human being, or if you are unable to decide whether the defendant was the actual killer or an aider or abet tor, you cannot find the special circumstance to be true unless you are satisfied beyond a reasonable doubt that the defendant with the intent to kill aided, abetted, counseled, commanded, induced, solicited, requested, or assisted any actor in the commission of the murder in the first degree, or with reckless indifference to human life and as a major participant, aided, abetted, coun seled, commanded, induced, solicited, requested, or as sisted in the commission of the crime of robbery (Penal Code s. 190, 2(1) (17) crime) which resulted in the death of a human being, namely Ramon Flores.

[J.A. 13]

* * * * *


CALJIC 8.81.17 (1991 Revision)
(Penal Code s. 190.2(a) (17))


To find that the special circumstance, referred to in these instructions as murder in the commission of rob bery is true, it must be proved:

1. The murder was committed while the de fendant was engaged in the commission or attempted commission of a robbery; or

2. The murder was committed in order to carry out or advance the commission of the crime of robbery or to facilitate the escape therefrom or to avoid detection. In other words, the special circum stance referred to in these instructions is not estab lished if the robbery was merely incidental to the com mission of the murder.

[J.A. 14]

* * * * *


CALJIC 9.40.1 (1991 New)


For the purposes of determining whether a person is guilty as an aider and abettor to robbery, the commis sion of the crime of robbery is not confined to a fixed place or a limited period of time and continues so long as the stolen property is being carried away to a place of temporary safety.

[J.A. 21]

* * * * *



For purposes of determining whether the defendant is guilty of robbery felony murder[,] the robbery continues until the robber has reached a position of temporary safety, with or without the stolen property.

[J.A. 23]

* * * * *


CALJIC 9.44 (1991 Revision)


For the purposes of determining whether an unlawful killing has occurred during the commission or attempted commission of a robbery, * * * [a] robbery is still in progress after the original taking of physical possession of the stolen property while the perpetrator is in posses sion of the stolen property and fleeing in an attempt to escape. * * *

A robbery is complete when the perpetrator eluded any pursuers, has reached a place of temporary safety, and is in unchallenged possession of the stolen property after having effected an escape with such property.

[J.A. 26]

* * * * *

Merits Stage Amicus Brief
Updated October 21, 2014