Danforth v. State

761 N.W.2d 493 (2009)

Stephen DANFORTH, Appellant,
v.
STATE of Minnesota, Respondent.

No. A04-1993.

Supreme Court of Minnesota.

February 26, 2009.

Lawrence Hammerling, Chief State Appellate Public Defender, Benjamin J. Butler, Assistant State Public Defender, St. Paul, MN, for appellant.

Lori Swanson, Attorney General, St. Paul, Minnesota; and Michael O. Freeman, Hennepin County Attorney, Patrick C. Diamond, Assistant County Attorney, Minneapolis, MN, for respondent.

OPINION

G. BARRY ANDERSON, Justice.

Appellant Stephen Danforth was charged with first-degree criminal sexual conduct involving a six-year-old boy, J.S. The boy was found incompetent to testify at trial, but the district court admitted into evidence a videotaped interview of J.S. conducted at a child advocacy center. The jury found Danforth guilty. The facts of this case are set forth at length in a decision by the court of appeals. State v. Danforth, 573 N.W.2d 369, 372 (Minn.App. 1997), rev. denied (Minn. Feb. 19, 1998). A recitation of the procedural events following Danforth's conviction can be found in our previous opinion, Danforth v. State, 718 N.W.2d 451, 454-55 (Minn.2006). In Danforth, we held that we were required to follow the standard for retroactivity of new rules of constitutional criminal procedure set by the United States Supreme Court in Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), and that Teague prohibited the retroactive application of the rule in Crawford v. Washington, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004), to Danforth's case. Danforth, 718 N.W.2d at 455, 460-61. The U.S. Supreme Court granted Danforth's writ of certiorari and reversed, holding that state courts, when reviewing state criminal convictions, are not bound by Teague and may choose the standard for deciding whether new rules of federal constitutional criminal procedure are retroactive. Danforth v. Minnesota, 552 U.S. ___, 128 S.Ct. 1029, 1046, 169 L.Ed.2d 859 (2008). We affirm.

On August 23, 1995, in a videotaped interview at CornerHouse, a child advocacy center that often interviews children who are alleged victims of sexual abuse, J.S. indicated that he had been sexually abused by Danforth. Danforth was arrested and charged with first-degree criminal sexual conduct. At Danforth's trial, the district court declared J.S. incompetent to testify, and admitted the videotape, finding that the videotape bore sufficient indicia of reliability in accordance with Minn.Stat. § 595.02, subd. 3 (1994). The jury viewed the videotape. The jury also heard testimony from J.S.'s sister, A.S., who testified that she had seen Danforth perform sexual acts on J.S. The jury convicted Danforth of first-degree criminal sexual conduct.

Eight years after Danforth's conviction, the Supreme Court in Crawford ruled that defendants have a right under the Sixth Amendment's Confrontation Clause to cross-examine testimonial witnesses. See Crawford, 541 U.S. at 68-69, 124 S.Ct. 1354. Crawford replaced the reliability standard, first announced in Ohio v. Roberts, 448 U.S. 56, 100 S.Ct. 2531, 65 L.Ed.2d 597 (1980), that had allowed admission of the videotape in Danforth's case. Crawford, 541 U.S. at 67-68, 124 S.Ct. 1354; Danforth, 718 N.W.2d at 454.

I.

The issue presented in this case is what standard we should use to decide whether new rules of federal constitutional criminal procedure will be applied retroactively. Appellant, who contends that his rights under the Confrontation Clause were violated by the admission at his trial of the videotaped statement of J.S., seeks to change the existing Teague standard and, utilizing a new standard, to retroactively apply to his case the rule announced in Crawford v. Washington, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004). We employ a de novo standard of review when reviewing questions of law. State v. Rodriguez, 754 N.W.2d 672, 678 (Minn. 2008). The issue of what standard we should use in deciding whether a new rule of federal constitutional criminal procedure should apply retroactively involves a question of law. Similarly, whether a defendant's Confrontation Clause rights have been violated also presents a question of law. State v. Krasky, 736 N.W.2d 636, 640 (Minn.2007). Consequently, we review the issues presented here under a de novo standard.

A.

Danforth argues we should announce a new standard for deciding whether to give retroactive effect to new federal constitutional rules of criminal procedure. He seeks either of two more relaxed standards, which, he contends, would enable him to challenge his conviction as based on an unconstitutional violation of his rights under the Sixth Amendment's Confrontation Clause. We begin with a review of the history of and principles behind our retroactivity standards.

From 1977 to 2004, we decided the retroactivity of new federal constitutional rules of criminal procedure under a standard modeled on two United States Supreme Court decisions: Linkletter v. Walker, 381 U.S. 618, 629, 85 S.Ct. 1731, 14 L.Ed.2d 601 (1965), and Stovall v. Denno, 388 U.S. 293, 297, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967). Under the so-called Linkletter-Stovall test, we decided whether to give retroactive effect to a particular decision based on (1) the purpose of the decision, (2) reliance on the prior rule of law, and (3) the effect upon the administration of justice of granting retroactive effect. See State v. Hamm, 423 N.W.2d 379, 386 (Minn.1988) (adopting the retroactivity criteria set forth in Linkletter and Stovall), superseded on other grounds by Minn. Const. art. I, § 4; State v. Olsen, 258 N.W.2d 898, 907 n. 15 (Minn.1977) (citing Linkletter, 381 U.S. at 629, 85 S.Ct. 1731).

The Linkletter-Stovall test was maligned by scholars and some Justices of the U.S. Supreme Court. In the four years after Linkletter was decided, the standard's application "produced strikingly divergent results" from the Supreme Court, depending on whether cases were subject to direct review, whether trials had commenced or not, whether tainted evidence had been introduced at trial, and other factors. Danforth, 128 S.Ct. at 1037 (citing Desist v. United States, 394 U.S. 244, 257, 89 S.Ct. 1030, 22 L.Ed.2d 248 (1969) (Harlan, J., dissenting)). Justice Harlan "reasonably questioned" whether Linkletter-Stovall's approach to retroactivity was producing decisions of a court of law or that of a super-legislature. Id. Commentators had "a veritable field day" with Linkletter-Stovall, with "much of the discussion [being] `more than mildly negative.'" Francis X. Beytagh, Ten Years of Non-Retroactivity: A Critique and a Proposal, 61 Va. L.Rev. 1557, 1558 (1975).

In 1989, the Supreme Court answered Justice Harlan's concerns by adopting a new approach to retroactivity for cases on collateral review.^[1] The Court decided a new rule would not normally be retroactively applied to a defendant's case once the defendant's case had become final. Teague v. Lane, 489 U.S. 288, 311, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989). The Teague standard allowed retroactive effect of a new rule in cases that had become final only under two exceptions: (1) when the rule places certain specific conduct beyond the power of the criminal law-making authority to proscribe, or (2) when the rule is a "watershed" rule of criminal procedure, and is a rule without which the likelihood of an accurate conviction would be seriously diminished. Id. at 311-12, 109 S.Ct. 1060. The Teague methodology later became the Court's majority approach to deciding retroactivity. Penry v. Lynaugh, 492 U.S. 302, 313, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989), overruled on other grounds by Atkins v. Virginia, 536 U.S. 304, 122 S.Ct. 2242, 153 L.Ed.2d 335 (2002).

We continued to follow Linkletter-Stovall until 2004. At that time, we stated that we were "compelled to follow the lead" of the United States Supreme Court when deciding the retroactivity of a new rule of federal constitutional criminal procedure, meaning we were obligated to follow the rulings in Teague and Griffith v. Kentucky, 479 U.S. 314, 107 S.Ct. 708, 93 L.Ed.2d 649 (1987). O'Meara v. State, 679 N.W.2d 334, 339 (Minn.2004). We held that in accordance with Griffith and Teague, "if a case is pending on direct review when a new rule of federal constitutional criminal procedure is announced, the defendant is entitled to benefit from that new rule. But if the defendant's conviction is already final at the time the new rule is announced, then the criminal defendant ordinarily may not avail himself of the new rule." O'Meara, 679 N.W.2d at 339-40 (footnotes omitted). We recognized the two exceptions to Teague's non-retroactivity standard: a new rule could be applied to convictions that have become final when (1) the rule "place[d] an entire category of primary conduct beyond the reach of the criminal law," or "prohibit[ed] imposition of a certain type of punishment for a class of defendants because of their status or offense"; or (2) the new rule was a "watershed" rule of criminal procedure that was "necessary to the fundamental fairness of the criminal proceeding." O'Meara, 679 N.W.2d. at 340 n. 3 quoting Sawyer v. Smith, 497 U.S. 227, 241-42, 110 S.Ct. 2822, 111 L.Ed.2d 193 (1990).

A year later, we again stated our belief that "the retroactivity principles of Teague. . . control," holding that an appellant whose conviction had become final before the new rule was announced in Blakely v. Washington, 542 U.S. 296, 124 S.Ct. 2531, 159 L.Ed.2d 403 (2004), could not invoke the new rule because it was not a watershed rule. State v. Houston, 702 N.W.2d 268, 270 (Minn.2005). Similarly, when Danforth sought the protection of the rule announced in Crawford v. Washington, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004), regarding admissibility of certain out-of-court testimonial statements under the Sixth Amendment's Confrontation Clause, we stated that we were required to apply Teague and held that Danforth was not entitled to retroactive application of the new Crawford rule. Danforth v. State, 718 N.W.2d 451, 455-56, 461 (Minn.2006).

But the U.S. Supreme Court, on review, held that its decisions do not limit the authority of state courts, when reviewing state criminal convictions, to fashion their own remedy in determining whether a new rule should be applied retroactively. Danforth v. Minnesota, 552 U.S. ___, 128 S.Ct. 1029, 1042, 169 L.Ed.2d 859 (2008). The Teague rule "was tailored to the unique context of federal habeas and therefore had no bearing on whether States could provide broader relief in their own postconviction proceedings." Danforth, 128 S.Ct. at 1039. States may follow Teague if they wish, but states may also choose to apply a broader retroactivity standard than stated in Teague. See id. at 1046-47. Thus, we may elect to retain the Teague standard even for non-federal cases, or we may determine retroactivity by a different standard of our choosing.

Teague was based on two primary policy interests: first, comity, meaning a policy against excessive interference by federal habeas courts in state criminal convictions that had become final, and, second, finality, meaning the need to bring criminal cases to a close. Danforth, 552 U.S. at ___, 128 S.Ct. at 1040-41; Teague, 489 U.S. at 306, 109 S.Ct. 1060. States that adopted Teague by choice did so primarily based on the important policy interest in finality. See Mary C. Hutton, Retroactivity in the States: The Impact of Teague v. Lane on State Postconviction Remedies, 44 Ala. L.Rev. 421, 458-59, 461-62 (1993). Some states have also valued uniformity between state and federal rules. See, e.g., State v. Slemmer, 170 Ariz. 174, 823 P.2d 41, 49 (1991) (stating that retroactivity law was "complex enough without requiring counsel and trial judges to apply different retroactivity rules, depending on whether the substantive decision is grounded on state or federal constitutional principles—especially when many decisions are grounded on both"); People v. Bradbury, 68 P.3d 494, 498 (Colo.Ct.App.2002) (seeing no compelling reason to depart from state's consistent policy of following lead of United States Supreme Court).

Some states have found Teague too narrow or strict, or out of place where a state court is reviewing its own convictions. See, e.g., Colwell v. State, 118 Nev. 807, 59 P.3d 463, 471 (2002) (adopting modified version of Teague). Since the Danforth ruling freed states to fashion their own retroactivity standards, only one state appears to have considered the question, and that state opted to retain Teague. See Ex Parte Lave, 257 S.W.3d 235 (Tex.Crim. App.2008) (concluding that "[a]lthough not required by the United States Supreme Court to do so, we adhere to our retroactivity analysis . . . that Crawford does not apply retroactively to cases on collateral review in Texas state courts.").

Danforth argues that we should abandon Teague. Danforth would have us either return to Linkletter-Stovall or adopt the approach outlined in Colwell. We elect to retain Teague. While we acknowledge that one of the policy concerns underlying Teague—that federal habeas courts not excessively interfere with state courts—is absent when a state court is reviewing state convictions, we continue to share the other policy concern behind Teague, which is the finality of convictions. Finality of state convictions is a matter that States are "free to evaluate, and weigh the importance of." Danforth, 552 U.S. at ___, 128 S.Ct. at 1041. We note, for example, an observation about finality that was quoted in Teague:

Application of constitutional rules not in existence at the time a conviction became final seriously undermines the principle of finality which is essential to the operation of our criminal justice system. Without finality, the criminal law is deprived of much of its deterrent effect. The fact that life and liberty are at stake in criminal prosecutions "shows only that `conventional notions of finality' should not have as much place in criminal as in civil litigation, not that they should have none."

Teague, 489 U.S. at 309, 109 S.Ct. 1060 (emphasis in original) (quoting Henry J. Friendly, Is Innocence Irrelevant? Collateral Attacks on Criminal Judgments, 38 U. Chi. L.Rev. 142, 150 (1970)).

We also note with favor the comments of Justice Harlan, issued in one of the cases that eventually led to Teague:

A rule of law that fails to take account of these finality interests would do more than subvert the criminal process itself. It would also seriously distort the very limited resources society has allocated to the criminal process. While men languish in jail, not uncommonly for over a year, awaiting a first trial on their guilt or innocence, it is not easy to justify expending substantial quantities of the time and energies of judges, prosecutors, and defense lawyers litigating the validity under present law of criminal convictions that were perfectly free from error when made final. This drain on society's resources is compounded by the fact that issuance of the habeas writ compels a State that wishes to continue enforcing its laws against the successful petitioner to relitigate facts buried in the remote past through presentation of witnesses whose memories of the relevant events often have dimmed. This very act of trying stale facts may well, ironically, produce a second trial no more reliable as a matter of getting at the truth than the first.

Mackey v. United States, 401 U.S. 667, 691, 91 S.Ct. 1160, 28 L.Ed.2d 404 (1971) (separate opinion of Harlan., J.) (citations omitted).

States have recognized this interest. The Florida Supreme Court summed up the concern this way:

The importance of finality in any justice system, including the criminal justice system, cannot be understated. It has long been recognized that, for several reasons, litigation must, at some point, come to an end. In terms of the availability of judicial resources, cases must eventually become final simply to allow effective appellate review of other cases. There is no evidence that subsequent collateral review is generally better than contemporaneous appellate review for ensuring that a conviction or sentence is just. Moreover, an absence of finality casts a cloud of tentativeness over the criminal justice system, benefiting neither the person convicted nor society as a whole.

Witt v. State, 387 So.2d 922, 925 (Fla.1980).

Teague may not be a perfect rule, but we believe it is preferable to the alternatives. Teague provides a bright line rule on the issue of when relief is to be retroactive. Defendants and parties know where that line is drawn—when a "judgment of conviction has been rendered, the availability of appeal exhausted, and the time for a petition for certiorari [has] elapsed or a petition for certiorari [has been filed and] finally denied." O'Meara v. State, 679 N.W.2d 334, 339 (Minn.2004) (quoting Griffith, 479 U.S. at 321, n. 6, 107 S.Ct. 708).

We see more disadvantages than benefits in returning to our previous retroactivity standard based on Linkletter-Stovall. Were we to return to the balancing test, we would likely face the harms described by Justice Harlan, as the balancing test could have the effect of reopening numerous criminal cases as defendants seek postconviction relief under the retroactive application of various "new" rules. For example, if we were to apply Crawford retroactively under the Linkletter-Stovall test, something we do not decide today, the criminal justice system would be burdened with litigating the validity of convictions, like Danforth's conviction, that were final before Crawford was announced and were perfectly free from error when finally decided. See State v. Rottelo, 2003 WL 21321400, at *4 (Minn.App.2003) (concluding, under the pre-Crawford standard, that videotape and child's statements admitted by the district court possessed sufficient reliability to be admissible). In addition, limited judicial resources may be expended litigating the convictions of defendants who never challenged the admission of a videotaped statement. See In re Welfare of C.L.W, 2005 WL 2207021, at *1 (Minn. App.2005). Consequently, we reject Danforth's argument that we should return to Linkletter-Stovall.

We choose not to adopt Colwell— Nevada's modified version of Teague—out of concern that it would only lead to the same problems that accompanied the Linkletter-Stovall standard. Colwell widened Teague's first exception, which is that a new rule is given retroactive effect if the rule establishes that it is unconstitutional to proscribe certain "primary, private individual conduct" as criminal. 59 P.3d at 472. Colwell said other conduct beyond primary, private individual conduct might also be protected from criminalization and warrant retroactive relief. Id. As to the second Teague exception, Colwell eliminated the bright-line requirement that the new criminal procedure rule have "watershed" or "bedrock" significance, instead declaring that "if accuracy is seriously diminished without the rule, the rule is significant enough to warrant retroactive application." Id. We note that no other state has followed Nevada's lead and adopted this framework. The Colwell standard would expand the retroactive application of new rules of constitutional procedure to cases where the absence of the new rule seriously diminished the accuracy of the trial but did not affect the fundamental fairness of the criminal proceeding. But relitigating cases in which a fundamentally fair trial has been held seriously distorts the allocation of very limited resources available to the criminal justice system. This is contrary to the well-established principle that a defendant is entitled to a fair trial, not a perfect trial. And the Colwell formula does not address one of the key failings of the Linkletter-Stovall test—a "serious" diminishment in "accuracy" as a trigger to retroactivity is not only not a bright-line standard, it invites the type of inconsistent and troubling results that so bedeviled the Linkletter-Stovall multi-factor test.

We are mindful of the criticism, voiced by the Nevada Supreme Court among others, that the Teague rule has been applied so strictly by the United States Supreme Court "that decisions defining a constitutional safeguard rarely merit application on collateral review." Colwell, 59 P.3d at 471. But Teague allows for the possibility that eventually there may be a new rule that is critical to fundamental fairness. That there have been no "watershed" rules announced in the 19 years since Teague does not mean Teague is unyielding or unworkable. It simply means that such cases are rare. Still, even as we formally adopt the Teague standard of our own volition, we are not bound by the U.S. Supreme Court's determination of fundamental fairness. Rather, we will independently review cases to determine whether they meet our understanding of fundamental fairness.

Affirmed.

Dissenting, ANDERSON, PAUL H., and PAGE, JJ.

PAUL H. ANDERSON, Justice.

I respectfully dissent. I would not adopt Teague in total, rather I would, as the Nevada Supreme Court has done, adopt the basic approach set forth in Teague but with some significant qualifications. See Colwell v. State, 118 Nev. 807, 59 P.3d 463 (2002).

The majority correctly states that in Danforth v. Minnesota, 552 U.S. ___, 128 S.Ct. 1029, 169 L.Ed.2d 859 (2008), the United States Supreme Court held that its decisions do not limit our authority as a state supreme court to fashion our own retroactivity rule when reviewing state criminal convictions. The Court took this position recognizing that the rule it fashioned in Teague "was tailored to the unique context of federal habeas, and therefore had no bearing on whether states could provide broader relief in their own post-conviction proceedings." Id. at 1039. In essence, the Court held that we may choose to adopt a retroactivity standard broader than the one articulated by the Court in Teague. We should do just that.

Certain aspects of the "approach to retroactivity set forth in Teague [are] sound in principle." Colwell, 59 P.3d at 471. Teague addresses a valid concern that the "finality of convictions not be unduly disturbed." Colwell, 59 P.3d at 471. Teague also attempts to provide a clear rule on when relief is to be retroactive. Thus, the Supreme Court in Teague properly attempted to address some of the deficiencies inherent in the Linkletter-Stovall test, a test that sometimes led to wide variations in results and drew criticism that the application of the test turned courts into legislatures. See Desist v. United States, 394 U.S. 244, 257, 89 S.Ct. 1030, 22 L.Ed.2d 248 (1969) (Harlan, J. dissenting).

The policy interests of finality and uniformity addressed by Teague are important, but I conclude that the Supreme Court has applied the Teague rule so narrowly and strictly that many cases involving constitutional safeguards that warrant collateral review have not or will not receive such review. Unlike the majority, I conclude that the fact there have been no "watershed" rules announced by the Court in the 19 years since it decided Teague sends a signal that the Teague test as applied by the Court is "unyielding or unworkable." I acknowledge that Teague may be more workable and applicable in the context of preventing excessive interference by federal habeas courts in their review of state court decisions. Nevertheless, when reviewing state criminal convictions, we should not limit our discretion in deciding retroactivity so narrowly. We should recognize the Court's acknowledgment that Teague is designed to address federal concerns and accept the Court's invitation to formulate our own rule to suit our own needs.

In Colwell, the Nevada Supreme Court broadened Teague's two exceptions. Teague's first exception gives retroactive effect to a new rule if that rule establishes that it is unconstitutional to prosecute certain "primary, private individual conduct" as criminal. Colwell, 59 P.3d at 472. Colwell acknowledged that some conduct beyond primary, private individual conduct might also be protected from criminalization and warrant retroactivity. Id. Colwell also rejected the "bright-line requirement" in Teague that results from its narrow and strict application of the "watershed" or "bedrock" significance requirement. Rather, the Nevada court declared that "if accuracy is seriously diminished without the rule, the rule is significant enough to warrant retroactive application." Id.

More specifically, the Nevada court defined its rule as follows:

Thus, consistent with the Teague framework, we will not apply a new constitutional rule of criminal procedure to finalized cases unless it falls within either of two exceptions. There is no bright-line rule for determining whether a rule is new, but there are basic guidelines to follow. As this court has stated, "When a decision merely interprets and clarifies an existing rule . . . and does not announce an altogether new rule of law, the court's interpretation is merely a restatement of existing law." Similarly, a decision is not new if "it has simply applied a well-established constitutional principle to govern a case which is closely analogous to those which have been previously considered in the prior case law." We consider too sweeping the proposition, noted above, that a rule is new whenever any other reasonable interpretation of prior law was possible. However, a rule is new, for example, when the decision announcing it overrules precedent, "or disapprove[s] a practice this Court has arguably sanctioned in prior cases, or overturn[s] a longstanding practice that lower courts had uniformly approved."

When a rule is new, it will still apply retroactively in two instances: (1) if the rule establishes that it is unconstitutional to proscribe certain conduct as criminal or to impose a type of punishment on certain defendants because of their status or offense; or (2) if it establishes a procedure without which the likelihood of an accurate conviction is seriously diminished.

Id. at 471-72 (internal citations omitted).

The majority rejects Nevada's modified version of Teague on the ground that such a standard will lead to the same problems that led to the rejection of the Linkletter-Stovall rule. Further, the majority expresses concern about distorting the allocation of "very limited resources available to the criminal justice system" and asserts that the Nevada rule improperly expands retroactive application of new rules of constitutional procedure to cases "where the absence of the new rules seriously diminishes the accuracy of the trial but did not affect the fundamental fairness of the criminal proceeding." (Emphasis added.) I do not share these concerns. Moreover, I fail to see how a trial that is conducted under a rule that seriously diminishes the accuracy of the trial is a "fundamentally fair trial." As the majority says, although a defendant may not be entitled to a perfect trial, he is entitled to a fair trial.

Thus, I disagree with the majority's decision to adopt Teague in total because the rule is too narrow and strict in its application. Rather, I would adopt as a template many of the sound principles embodied in the framework of the Teague test but, as the Nevada Supreme Court did, "reserve our prerogatives to define and determine within this framework whether a rule is new and whether it falls within the two exceptions to non-retroactivity." Colwell, 59 P.3d at 471. I agree with the Nevada Court that "adaptation of the approach taken in Teague and its progeny" will provide us "with a fair and straightforward framework for determining retroactivity." Id. at 472. Adopting a modified Teague standard would enable us, as the Oregon Supreme Court has said, to be "free to choose the degree of retroactivity or prospectivity which we believe appropriate to the particular rule under consideration, so long as we give federal constitutional rights at least as broad a scope as the United States Supreme Court requires." State v. Fair, 263 Or. 383, 502 P.2d 1150, 1152 (1972).

PAGE, J. (dissenting).

I join in the dissent of Justice Paul H. Anderson.

NOTES

[1] Before it adopted a new approach to retroactivity for cases on collateral review, which is our concern here, the Supreme Court first rejected the Linkletter-Stovall standard for cases pending on direct review at the time a new rule is announced. Griffith v. Kentucky, 479 U.S. 314, 322-23, 107 S.Ct. 708, 93 L.Ed.2d 649 (1987).