McGautha v. California (402 U.S. 183)/Dissent Brennan

Mr. Justice BRENNAN, with whom Mr. Justice DOUGLAS and Mr. Justice MARSHALL join, dissenting.

These cases test the viability of principles whose roots draw strength from the very core of the Due Process Clause. The question that petitioners present for our decision is whether the rule of law, basic to our society and binding upon the States by virtue of the Due Process Clause of the Fourteenth Amendment, is fundamentally inconsistent with capital sentencing procedures that are purposely constructed to allow the maximum possible variation from one case to the next, and provide no mechanism to prevent that consciously maximized variation from reflecting merely random or arbitrary choice. The Court does not, however, come to grips with that fundamental question. Instead, the Court misapprehends petitioners' argument and deals with the cases as if petitioners contend that due process requires capital sentencing to be carried out under predetermined standards so precise as to be capable of purely mechanical application, entirely eliminating any vestiges of flexibility or discretion in their use. This misapprehended question is then treated in the context of the Court's assumption that the legislatures of Ohio and California are incompetent to express with clarity the bases upon which they have determined that some persons guilty of some crimes should be killed, while others should live-an assumption that, significantly, finds no support in the arguments made by those States in these cases. With the issue so polarized, the Court is led to conclude that the rule of law and the power of the States to kill are in irreconcilable conflict. This conflict the Court resolves in favor of the States' power to kill.

In my view the Court errs at all points from its premises to its conclusions. Unlike the Court, I do not believe that the legislatures of the 50 States are so devoid of wisdom and the power of rational thought that they are unable to face the problem of capital punishment directly, and to determine for themselves the criteria under which convicted capital felons should be chosen to live or die. We are thus not, in my view, faced by the dilemma perceived by the Court, for cases in this Court have for almost a century and a half approved a multiplicity of imaginative procedures designed by the state and federal legislatures to assure evenhanded treatment and ultimate legislative control regarding matters that the legislatures have deemed either too complex or otherwise inapposite for regulation under predetermined rules capable of automatic application in every case. Finally, even if I shared the Court's view that the rule of law and the power of the States to kill are in irreconcilable conflict, I would have no hesitation in concluding that the rule of law must prevail.

Except where it incorporates specific substantive constitutional guarantees against state infringement, the Due Process Clause of the Fourteenth Amendment does not limit the power of the States to choose among competing social and economic theories in the ordering of life within their respective jurisdictions. But it does require that, if state power is to be exerted, these choices must be made by a responsible organ of state government. For if they are not, the very best that may be hoped for is that state power will be exercised, not upon the basis of any social choice made by the people of the State, but instead merely on the basis of social choices made at the whim of the particular state official wielding the power. If there is no effective supervision of this process to insure consistency of decision, it can amount to nothing more than government by whim. But ours has been 'termed a government of laws, and not of men.' Marbury v. Madison, 1 Cranch 137, 163, 2 L.Ed. 60 (1803). Government by whim is the very antithesis of due process.

It is not a mere historical accident that '(t)he history of liberty has largely been the history of observance of procedural safeguards.' McNabb v. United States, 318 U.S. 332, 347, 63 S.Ct. 608, 616, 87 L.Ed. 819 (1943) (Frankfurter, J.). The range of permissible state choice among competing social and economic theories is so broad that almost any arbitrary or otherwise impermissible discrimination among individuals may mask itself as nothing more than such a permissible exercise of choice unless procedures are devised which adequately insure that the relevant choice is actually made. Such procedures may take a variety of forms. The decisionmaker may be provided with a set of guidelines to apply in rendering judgment. His decision may be required to rest upon the presence or absence of specific factors. If the legislature concludes that the range of variation to be dealt with precludes adequate treatment under inflexible, predetermined standards it may adopt more imaginative procedures. The specificity of standards may be relaxed, directing the decisionmaker's attention to the basic policy determinations underlying the statute without binding his action with regard to matters of important but unforeseen detail. He may be instructed to consider a list of factors-either illustrative or exhaustive intended to illuminate the question presented without setting a fixed balance. The process may draw upon the genius of the common law, and direct itself toward the refinement of understanding through case-by-case development. In such cases decision may be left almost entirely in the hands of the body to which it is delegated, with ultimate legislative supervision on questions of basic policy afforded by requiring the decisionmakers to explain their actions, and evenhanded treatment enhanced by requiring disputed factual issues to be resolved and providing for some form of subsequent review. Creative legislatures may devise yet other procedures. Depending upon the nature and importance of the issues to be decided, the kind of tribunal rendering judgment, the number and frequency of decisions to be made, and the number of separate tribunals involved in the process, these techniques may be applied singly or in combination.

It is of critical importance in the present cases to emphasize that we are not called upon to determine the adequacy or inadequacy of any particular legislative procedure designed to give rationality to the capital sentencing process. For the plain fact is that the legislatures of California and Ohio, whence come these cases, have sought no solution at all. We are not presented with a State's attempt to provide standards, attacked as impermissible or inadequate. We are not presented with a legislative attempt to draw wisdom from experience through a process looking toward growth in understanding through the accumulation of a variety of experiences. We are not presented with the slightest attempt to bring the power of reason to bear on the considerations relevant to capital sentencing. We are faced with nothing more than stark legislative abdication. Not once in the history of this Court, until today, have we sustained against a due process challenge such an unguided, unbridled, unreviewable exercise of naked power. Almost a century ago, we found an almost identical California procedure constitutionally inadequate to license a laundry. Yick Wo v. Hopkins, 118 U.S. 356, 366-367, 369 370, 6 S.Ct. 1064, 1068-1069, 1070-1071, 30 L.Ed. 220 (1886). Today we held it adequate to license a life. I would reverse petitioners' sentences of death.

* 'Our scheme of ordered liberty is based, like the common law, on enlightened and uniformly applied legal principle, not on ad hoc notions of what is right or wrong in a particular case.' J. Harlan, Thoughts at a Dedication: Keeping the Judicial Function in Balance, in The Evolution of a Judicial Philosophy 289, 291-292 (D. Shapiro ed., 1969). The dangers inherent in any grant of governmental power without procedural safeguards upon its exercise were known to English law long long before the Constitution was established. See, e.g., 8 How.St.Tr. 55-58, n. The principle that our Government shall be of laws and not of men is so strongly woven into our constitutional fabric that it has found recognition in not just one but several provisions of the Constitution. And this principle has been central to the decisions of this Court giving content to the Due Process Clause. As we said in Hurtado v. California, 110 U.S. 516, 535-536, 4 S.Ct. 111, 120-121, 28 L.Ed. 232 (1884):

'(I)t is not to be supposed that * *  * the amendment      prescribing due process of law is too vague and indefinite to operate as a practical restraint. * *  * Law is      something more than mere will exerted as an act of power. It     must be not a special rule for a particular person or a      particular case, but *  *  * 'the general law *  *  * ' so 'that      every citizen shall hold his life, liberty, property and      immunities under the protection of the general rules which      govern society,' and thus excluding, as not due process of      law, acts of attainder, bills of pains and penalties, acts of      confiscation *  *  * and other similar special, partial and      arbitrary exertions of power under the forms of legislation. Arbitrary power, enforcing its edicts to the injury of the     persons and property of its subjects, is not law, whether      manifested as the decree of a personal monarch or of an      impersonal multitude.'

The principal function of the Due Process Clause is to insure that state power is exercised only pursuant to procedures adequate to vindicate individual rights. While we have, on rare occasions, held that due process requires specific procedural devices not explicitly commanded by the Bill of Rights, we have generally either indicated one acceptable procedure and left the States free to device others, or else merely ruled upon the validity or invalidity of a particular procedure without attempting to limit or even guide state choice of procedural mechanisms beyond stating the obvious proposition that inadequate mechanisms may not be employed. Several principles, however, have until today been consistently employed to guide determinations of the adequacy of any given state procedure. 'When the Government exacts * *  * much, the importance of fair, evenhanded, and uniform decisionmaking is obviously intensified.' Gillette v. United States, 401 U.S. 437, 455, 91 S.Ct. 828, 839, 28 L.Ed.2d 168 (1971). Procedures adequate to determine a welfare claim may not suffice to try a felony charge. Compare Goldberg v. Kelly, 397 U.S. 254, 270-271, 90 S.Ct. 1011, 1021-1022, 25 L.Ed.2d 287 (1970), with Gideon v. Wainwright, 372 U.S. 335, 83 S.Ct. 792, 9 L.Ed.2d 799 (1963). Second, even where the only rights to be adjudicated are those created and protected by state law, due process requires that state procedures be adequate to allow all those concerned a fair hearing of their state-law claims. Boddie v. Connecticut, 401 U.S. 371, 91 S.Ct. 780, 28 L.Ed.2d 113 (1971) Covey v. Town of Somers, 351 U.S. 141, 76 S.Ct. 724, 100 L.Ed. 1021 (1956); Mullane v. Central Hanover Trust Co., 339 U.S. 306, 70 S.Ct. 652, 94 L.Ed. 865 (1950). Third, where federally protected rights are involved, due process commands not only that state procedure be adequate to assure a fair hearing of federal claims, In re Gault, 387 U.S. 1, 87 S.Ct. 1428, 18 L.Ed.2d 527 (1967), but also that it provide adequate opportunity for review of those federal claims where such review is otherwise available. Goldberg v. Kelly, 397 U.S., at 271, 90 S.Ct., at 1022; Boykin v. Alabama, 395 U.S. 238, 242-244, 89 S.Ct. 1709, 1711-1713, 23 L.Ed.2d 274 (1969); Jackson v. Denno, 378 U.S. 368, 387, 84 S.Ct. 1774, 1786, 12 L.Ed.2d 908 (1964); cf. North Carolina v. Pearce, 395 U.S. 711, 725-726, 89 S.Ct. 2072, 2080-2081, 23 L.Ed.2d 656 (1969); In re Murchison, 349 U.S. 133, 136, 75 S.Ct. 623, 625, 99 L.Ed. 942 (1955). Finally, and closely related to the previous point, due process requires that procedures for the exercise of state power be structured in such a way that, ultimately at least, fundamental choices among competing state policies are resolved by a responsible organ of state government. Louisiana v. United States, 380 U.S. 145, 152-153, 85 S.Ct. 817, 821-822, 13 L.Ed.2d 709 (1965) (BLACK, J.); FCC v. RCA Communications, Inc., 346 U.S. 86, 90, 73 S.Ct. 998, 1002, 97 L.Ed. 1470 (1953); Niemotko v. Maryland, 340 U.S. 268, 71 S.Ct. 325, 95 L.Ed. 267 (1951); United States v. Rock Royal Co-op, 307 U.S. 533, 574, 575, 59 S.Ct. 993, 1013, 83 L.Ed. 1446 (1939); Currin v. Wallace, 306 U.S. 1, 15, 59 S.Ct. 379, 386-387, 83 L.Ed. 441 (1939); Lovell v. Griffin, 303 U.S. 444, 58 S.Ct. 666, 82 L.Ed. 949 (1938); Browning v. Hooper, 269 U.S. 396, 405-406, 46 S.Ct. 141, 143, 70 L.Ed. 330 (1926); McKinley v. United States, 249 U.S. 397, 399, 39 S.Ct. 324, 63 L.Ed. 668 (1919); Eubank v. Richmond, 226 U.S. 137, 143-144, 33 S.Ct. 76, 77, 57 L.Ed. 156 (1912); Yick Wo v. Hopkins, 118 U.S., at 366-367, 369-370, 6 S.Ct., at 1068-1069, 1070-1071. The damage that today's holding, if followed, would do to our constitutional fabric can only be understood from a closer examination of our cases than is contained in the Court's opinion. I therefore turn to those cases.

* Analysis may usefully begin with this Court's cases applying what has come to be known as the 'void-for-vagueness' doctrine. It is sometimes suggested that in holding a statute void for vagueness, this Court is merely applying one of two separate doctrines: first, that a criminal statute must give fair notice of the conduct that it forbids, e.g., Lanzetta v. New Jersey, 306 U.S. 451, 59 S.Ct. 618, 83 L.Ed. 888 (1939); Connally v. General Construction Co., 269 U.S. 385, 391, 46 S.Ct. 126, 127, 70 L.Ed. 322 (1926); and second, that a statute may not constitutionally be enforced if it indiscriminately sweeps within its ambit conduct that may not be the subject of criminal sanctions as well as conduct that may. E.g., Baggett v. Bullitt, 377 U.S. 360, 84 S.Ct. 1316, 12 L.Ed.2d 377 (1964); Dombrowski v. Pfister, 380 U.S. 479, 492-496, 85 S.Ct. 1116, 1124-1126, 14 L.Ed.2d 22 (1965). To this is often added the observation that both doctrines apply with particular vigor to state regulation of conduct at or near the boundaries of the First Amendment. See United States v. National Dairy Corp., 372 U.S. 29, 36, 83 S.Ct. 594, 599-600, 9 L.Ed.2d 561 (1963); Smith v. California, 361 U.S. 147, 150-152, 80 S.Ct. 215, 217-218, 4 L.Ed.2d 205 (1959). But unless it be assumed that our decisions in such matters have shown an almost unparalleled inconsistency, these factors may not be taken as more than a partial explanation of the doctrine.

To begin with, we have never treated claims of unconstitutional statutory vagueness in terms of the statute as written or as construed prior to the time of the conduct in question. Instead, we have invariably dealt with the statute as glossed by the courts below at the time of decision here. E.g., Giaccio v. Pennsylvania, 382 U.S. 399, 86 S.Ct. 518, 15 L.Ed.2d 447 (1966); Winters v. New York, 333 U.S. 507, 68 S.Ct. 665, 92 L.Ed. 840 (1948); Cox v. New Hampshire, 312 U.S. 569, 61 S.Ct. 762, 85 L.Ed. 1049 (1941). In Musser v. Utah, 333 U.S. 95, 68 S.Ct. 397, 92 L.Ed. 562 (1948), we even remanded a criminal case to the Utah Supreme Court for a construction of the statute so that its possible vagueness could be analyzed. In dealing with vagueness attacks on federal statutes, we have not hesitated to construe the statute to avoid vagueness problems and, having so construed it, apply it to the case at hand. See United States v. Vuitch, 402 U.S. 62, 91 S.Ct. 1294, 28 L.Ed.2d 601 (1971); Dennis v. United States, 341 U.S. 494, 502, 71 S.Ct. 857, 863, 95 L.Ed. 1137 (1951); Kay v. United States, 303 U.S. 1, 58 S.Ct. 468, 82 L.Ed. 607 (1938). If the vagueness doctrine were fundamentally premised upon a concept of fair notice, such treatment would simply make no sense: a citizen cannot be expected to foresee subsequent construction of a statute by this or any other court. See Freund, The Supreme Court and Civil Liberties, 4 Vand.L.Rev. 533, 540-542 (1951). But if, as I believe, the doctrine of vagueness is premised upon the fundamental notion that due process requires governments to make explicit their choices among competing social policies, see infra, at 259-265, the inconsistency between theory and practice disappears. Of course such a choice, once made, is not irrevocable: statutes may be amended and statutory construction overruled. Nevertheless, an explicit state choice among possible statutory constructions substantially reduces the likelihood that subsequent convictions under the statute will be based on impermissible factors. It also renders more effective the available mechanisms for judicial review, by increasing the likelihood that impermissible factors, if relied upon, will be discernible from the record. Thus in Thompson v. Louisville, 362 U.S. 199, 80 S.Ct. 624, 4 L.Ed.2d 654 (1960), we were faced with the application of a specific vagrancy statute to conduct-dancing in a public bar-that there is no reason to believe could not have been constitutionally prohibited had the State chosen to do so. We were, however, able to examine the record and conclude that there was in fact no evidence that could support a conviction under the statute. Cf. Bachellar v. Maryland, 397 U.S. 564, 90 S.Ct. 1312, 25 L.Ed.2d 570 (1970) (impossible to determine whether verdict rested upon permissible or impermissible grounds).

Second, in dealing with statutes that are unconstitutionally overbroad, we have consistently indicated that 'once an acceptable limiting construction is obtained, (such a statute) may be applied to conduct occurring prior to the construction, provided such application affords fair warning to the defendants.' Dombrowski v. Pfister, 380 U.S., at 491 n. 7, 85 S.Ct., at 1123 (citations omitted); see, e.g., Poulos v. New Hampshire, 345 U.S. 395, 73 S.Ct. 760, 97 L.Ed. 1105 (1953). That is, an unconstitutionally overbroad statute may not be enforced at all until an acceptable construction has been obtained, e.g., Thornhill v. Alabama, 310 U.S. 88, 60 S.Ct. 736, 84 L.Ed. 1093 (1940); but once such a construction has been made, the statute as construed may be applied to conduct occurring prior to the limiting construction. If notice and overbreadth were the only components of the vagueness doctrine, this treatment would, once again, be inexplicable. So far as notice is concerned, one who has engaged in certain conduct prior to the limiting construction of an overbroad statute has obviously not received from that construction any warning that would have enabled him to keep his conduct within the bounds of law. Similarly, if adequate notice has in fact been given by an overbroad statute that certain conduct was criminally punishable, it is hard to see how the doctrine of overbreadth is furthered by forbidding the State, on the one hand, to punish that conduct so long as an acceptable limiting construction has not been obtained, but permitting it to punish the same, prior conduct once the statute has been acceptably construed. Once again, however, our actions are not at all inexplicable if examined in the terms articulated here. Once an acceptable limiting construction has in fact been obtained, there is by that very fact an assurance that a responsible organ of state power has made an explicit choice among possible alternative policies: for it should not be forgotten that the States possess constitutional power to make criminal much conduct that they may not wish to forbid, or may even desire to encourage. If a vague or overbroad statute is applied before it has been acceptably construed, there remains the danger that an individual whose conduct is admittedly clearly within the scope of the statute on its face will be punished for actions which in fact the State does not desire to make generally punishable-conduct which, if engaged in by another person, would not be subject to criminal liability. Shuttlesworth v. Birmingham, 382 U.S. 87, 91-92, 86 S.Ct. 211, 213-214, 15 L.Ed.2d 176 (1965). Allowing a vague or overbroad statute to be enforced if, and only if, an acceptable construction has been obtained forces the State to make explicit its social choices and prevents discrimination through the application of one policy to one person and another policy to others.

Particularly relevant to the present case is our decision in Giaccio v. Pennsylvania, 382 U.S. 399, 86 S.Ct. 518, 15 L.Ed.2d 447 (1966). That case involved a statute whereby Pennsylvania attempted to mitigate the harshness of its common-law rule requiring criminal defendants to pay the costs of prosecution in all cases by committing the matter to the discretion of the jury in cases where the defendant was found not guilty. Two members of this Court, concurring in the result, would have held that due process forbade the imposition of costs upon an acquitted defendant. 382 U.S., at 405, 86 S.Ct., at 522. We refused, however, to base our decision on that ground. In an opinion by my Brother BLACK, we said:

'We agree with the trial court * *  * that the 1860 Act is      invalid under the Due Process Clause because of vagueness and      the absence of any standards sufficient to enable defendants      to protect themselves against arbitrary and discriminatory      impositions of costs.

' * *  * It is established that a law fails to meet the      requirements of the Due Process Clause if it is so vague and      standardless that it leaves the public uncertain as to the      conduct it prohibits or leaves judges and jurors free to      decide, without any legally fixed standards, what is      prohibited and what is not in each particular case. This 1860     Pennsylvania Act contains no standards at all *  *  *. Certainly one of the basic purposes of the Due Process Clause     has always been to protect a person against having the Government impose burdens upon him except in accordance with      the valid laws of the land. Implicit in this constitutional     safeguard is the premise that the law must be one that      carries an understandable meaning with legal standards that      courts must enforce. * *  *

' * *  * The State contends that *  *  * state court      interpretations have provided standards and guides that cure      the *  *  * constitutional deficiencies. We do not agree. * *  *      In this case the trial judge instructed the jury that it      might place the costs of prosecution on the appellant, though      found not guilty of the crime charged, if the jury found that      'he has been guilty of some misconduct less than the offense      which is charged but nevertheless misconduct of some kind as      a result of which he should be required to pay some penalty      short of conviction (and) *  *  * his misconduct has given rise      to the prosecution.'

'It may possibly be that the trial court's charge comes     nearer to giving a guide to the jury than those that preceded      it, but it still falls short of the kind of legal standard      due process requires. * *  * ' 382 U.S., at 402-404, 86 S.Ct.,      at 520-521 (emphasis added) (citations omitted).

Several features of Giaccio are especially pertinent in the present context. First, there were no First Amendment implications in either the conduct charged or that in which Giaccio claimed to have engaged: the State's evidence was to the effect that Giaccio had wantonly discharged a firearm at another, in violation of Pa.Stat.Ann., Tit. 18, § 4716 (1963), and Giaccio's defense was that 'the firearm he had discharged was a starter pistol which only fired blanks.' 392 U.S., at 400, 86 S.Ct., at 519. Second, we were not presented with a defendant who had been convicted for conduct he could not have known was unlawful. Whether or not Giaccio's actions fell within § 4716, his conduct was unquestionably punishable under other state laws, e.g., Pa.Stat.Ann. Tit. 18, § 4406 (1963). Finally, it is worthy of note that in Giaccio two members of this Court explicitly sought to base the result upon the ground that, as a matter of substantive due process, the States were forbidden to impose the costs of prosecution upon an acquitted defendant. 382 U.S., at 405, 86 S.Ct., 522 (concurring opinions of Stewart and Fortas, JJ.). Yet we refused to place decision on any such ground. We held instead, consistently with our prior decisions, that the procedure for determining Giaccio's punishment lacked the safeguards against arbitrary action that are required by due process of law.

Our decisions applying the Due Process Clause through the doctrine of unconstitutional vagueness, then, lead to the following conclusions. First, the protection against arbitrary and discriminatory action embodied in the Due Process Clause requires that state power be exerted only through mechanisms that assure that fundamental choices among competing state policies be explicitly made by some responsible organ of the State. Second, the cases suggest that due process requires as well that state procedures for decision of questions that may have adverse consequences for an individual neither leave room for the deprivation sub silentio of the individual's federally protected rights nor unduly frustrate the federal judicial review provided for the vindication of those rights. This second point is explicitly made in a not unrelated line of cases, to which I now turn.

Whether through its own force or only through the application of other, specific constitutional guarantees, the Due Process Clause of the Fourteenth Amendment protects individuals from a narrow class of impermissible exertions of power by the States. As applied to the procedures whereby admittedly permissible state power is exerted, however, the Due Process Clause has consistently been given a wider scope. '(O)ur system of law has always endeavored to prevent even the probability of unfairness.' In re Murchison, 349 U.S. 133, 136, 75 S.Ct. 623, 625, 99 L.Ed. 942 (1955). Thus, we have never suggested that every judge who has been the target of contemptuous, personal attacks by litigants or their attorneys is incapable of rendering a fair decision on the merits of a contempt charge against such persons; but we have consistently held that, excepting only cases of urgent necessity, due process requires that contempt charges in such cases be heard by a different judge. Mayberry v. Pennsylvania, 400 U.S. 455, 91 S.Ct. 499, 27 L.Ed.2d 532 (1971); In re Murchison, supra. And in Tumey v. Ohio, 273 U.S. 510, 47 S.Ct. 437, 71 L.Ed. 749 (1927), we did not suggest that every judgment rendered by an official who had a financial stake in the outcome was ipso facto the product of bias. Proceeding from a directly contrary assumption, we nevertheless held that due process was violated by any 'procedure which would offer a possible temptation to the average man * *  * not to hold the balance nice, clear and true between the State and the accused.' Id., at 532, 47 S.Ct., at 444. In Jackson v. Denno, 378 U.S. 368, 84 S.Ct. 1774, 12 L.Ed.2d 908 (1964), one of the two grounds on which we struck down a New York procedure that required a jury to determine the voluntariness of a confession at the same time that it determined his guilt of the crime charged was that the procedure created an impermissible-and virtually unreviewable-risk that the jury would not be able to disregard a confession that it felt was both involuntary and true. Id., at 388-391, 84 S.Ct., at 1786-1787. Similarly, in a long line of cases beginning with Lovell v. Griffin, 303 U.S. 444, 58 S.Ct. 666, 82 L.Ed. 949 (1938), we have repeatedly held that due process is violated by state procedures for the administration of permit systems regulating the public exercise of First Amendment rights if the procedure allows a permit to be denied for impermissible reasons, whether or not an individual can actually demonstrate that he was denied a permit for activity which the State could not lawfully prohibit. And only recently, in Louisiana v. United States, 380 U.S. 145, 85 S.Ct. 817, 13 L.Ed.2d 709 (1965), we were faced with a state procedure for determining voting qualifications that, in the State's own words, vested 'discretion in the registrars of voters to determine the qualifications of applicants for registration,' but imposed 'no definite and objective standards upon registrars of voters for the administration of the interpretation test.' Id., at 152, 85 S.Ct., at 822. After quoting, with apparent approval, an 1898 state criticism of a similar procedure on the ground that the 'arbitrary power, lodged with the registration officer, practically places his decision beyond the pale of judicial review,' Ibid., we noted and accepted the District Court's finding that 'Louisiana * *  * provides no effective method whereby arbitrary and capricious action by registrars of voters may be prevented or redressed.' Ibid. We continued:

'The applicant facing a registrar in Louisiana thus has been     compelled to leave his voting fate to that official's      uncontrolled power to determine whether the applicant's      understanding of the Federal or State Constitution is      satisfactory. * *  * The cherished right of people in a country like ours to vote cannot be      obliterated by the use of laws like this, which leave the      voting fate of a citizen to the passing whim or impulse of an      individual registrar. Many of our cases have pointed out the     invalidity of laws so completely devoid of standards and      restraints.' 380 U.S., at 152-153, 85 S.Ct., at 822.

On that basis we held the Louisiana procedure for determining the qualifications of prospective voters to be a denial of due process. Ibid.

Diverse as they are, these cases rest upon common ground. They all stand ineluctably for the proposition that due process requires more of the States than that they not exert state power in impermissible ways. Specifically, the rule of these cases is that state procedures are inadequate under the Due Process Clause unless they are designed to control arbitrary action and also to make meaningful the otherwise available mechanism for judicial review. We have elsewhere made this last point explicit. In Specht v. Patterson, 386 U.S. 605, 87 S.Ct. 1209, 18 L.Ed.2d 326 (1967), we held that due process in commitment proceedings, 'whether denominated civil or criminal,' id., at 608, 87 S.Ct., at 1211, requires 'findings adequate to make meaningful any appeal that is allowed.' Id., at 610, 87 S.Ct., at 1212; see Garner v. Louisiana, 368 U.S. 157, 173, 82 S.Ct. 248, 256, 7 L.Ed.2d 207 (1961). And in Jackson v. Denno, supra, the alternative ground on which we struck down a New York procedure for determining the voluntariness of a confession by submitting that question to the jury at the same time as the question of guilt was that the 'admixture of reliability and voluntariness in the considerations of the jury would itself entitle a defendant to further proceedings in any case in which the essential facts are disputed, for we cannot determine how the jury resolved these issues and will not assume that they were reliably and properly resolved against the accused.' 378 U.S., at 387, 84 S.Ct., at 1786 (emphasis added). In other words, due process forbids the States to adopt procedures that would defeat the institution of federal judicial review.

The depth to which these principles are embedded in the concept of due process is evidenced by the fact that we have, on occasion, applied them not merely to rule that a particular state procedure is or is not permissible under the Due Process Clause, but that a particular, specific procedure is required by due process. We have repeatedly held, for example, that a guilty plea and its inevitably attendant waivers of federally guaranteed rights are valid only if they represent a 'voluntary and intelligent choice' on the part of the defendant. North Carolina v. Alford, 400 U.S. 25, 31, 91 S.Ct. 160, 164, 27 L.Ed.2d 162 (1970). The validity of a guilty plea may be tested on federal habeas corpus, where facts outside the record may be pleaded and proved. Waley v. Johnston, 316 U.S. 101, 62 S.Ct. 964, 86 L.Ed. 1302 (1942). While recognizing the existence of such a remedy, we held in Boykin v. Alabama, 395 U.S. 238, 89 S.Ct. 1709, 23 L.Ed.2d 274 (1969), that due process requires a record 'adequate for any review that may be later sought,' id., at 244, 89 S.Ct., at 1712, and does not permit protection of the federally guaranteed rights to be relegated to 'collateral proceedings that seek to probe murky memories.' Ibid. Accordingly, we held that due process requires a State, in accepting a plea of guilty, to make a contemporaneous record adequate 'to show that (the defendant) had intelligently and knowingly pleaded guilty.' Id., at 241, 89 S.Ct., at 1711. And only last Term, in Goldberg v. Kelly, 397 U.S. 254, 90 S.Ct. 1011, 25 L.Ed.2d 287 (1970), we held that because a decision on the withdrawal of welfare benefits must 'rest solely on the legal rules and evidence adduced at the hearing,' id., at 271, 90 S.Ct., at 1022, due process requires that the decisionmaker 'demonstrate compliance with this elementary requirement' by 'stat(ing) the reasons for his determination and indicat(ing) the evidence he relied on.' Ibid.

In my view, the cases discussed above establish beyond peradventure the following propositions. First, due process of law requires the States to protect individuals against the arbitrary exercise of state power by assuring that the fundamental policy choices underlying any exercise of state power are explicitly articulated by some responsible organ of state government. Second, due process of law is denied by state procedural mechanisms that allow for the exercise of arbitrary power without providing any means whereby arbitrary action may be reviewed or corrected. Third, where federally protected rights are involved due process of law is denied by state procedures which render inefficacious the federal judicial machinery that has been established for the vindication of those rights. If there is any way in which these propositions must be qualified, it is only that in some circumstances the impossibility of certain procedures may be sufficient to permit state power to be exercised notwithstanding their absence. Cf. Carroll v. President and Commissioners, 393 U.S. 175, 182, 184-185, 89 s.Ct. 347, 352, 353-354, 21 L.Ed.2d 325 (1968). But the judgment that a procedural safeguard otherwise required by the Due Process Clause is impossible of application in particular circumstances is not one to be lightly made. This is all the more so when as in the present cases, the argument of impossibility is not made by the parties before us, but only by this Court. Before we conclude that capital sentencing is inevitably a matter of such complexity that it cannot be carried out in consonance with the fundamental requirements of due process, we should at the very least examine the mechanisms developed in not incomparable situations and previously approved by this Court. Therefore, before examing the specific capital sentencing procedures at issue in these cases in light of the Due Process Clause, I am compelled to discuss both the mechanisms available for the control of arbitrary action and the nature of the capital sentencing process.

A legislature that has determined that the State should kill some but not all of the persons whom it has convicted of certain crimes must inevitably determine how the State is to distinguish those who are to be killed from those who are not. Depending ultimately on the legislature's notion of wise penological policy, that distinction may be hard or easy to make. But capital sentencing is not the only difficult question with which legislatures have ever been faced. At least since Wayman v. Southard, 23 U.S. (10 Wheat.) 1, 6 L.Ed. 253 (1825), we have recognized that the Constitution does not prohibit Congress from dealing with such questions by delegating to others the responsibility for their determination. It is not my purpose to trace in detail either the sources and scope of the delegation doctrine or the extent to which it is applicable to the States through the Due Process Clause. It is sufficient to state that in my view, whatever the sources of the doctrine, its application to the States as a matter of due process is merely a reflection of the fundamental principles of due process already discussed: in my Brother Harlan's words, the delegation doctrine

'insures that the fundamental policy decisions in our society     will be made not by an appointed official but by the body      immediately responsible to the people (and) prevents judicial      review from becoming merely an exercise at large by providing      the courts with some measure against which to judge the      official action that has been challenged.' Arizona v.

California, 373 U.S. 546, 626, 83 S.Ct. 1468, 1511, 10     L.Ed.2d 542 (1963) (dissenting in part).

My intention here is merely to provide an admittedly brief sketch of the several mechanisms that Congress has employed to assure that even with regard to the most complex and intractable problems, delegation by Congress of the power to make law has been subject to controls that limit the possibility of arbitrary action and that assure that Congress retains the responsibility for ultimate decision of fundamental questions of national policy. With these mechanisms in mind, I intend briefly to discuss the considerations relevant to the problem of capital sentencing with an eye to the question whether it may responsibly be said that all of these mechanisms are impossible of application by the States to the capital sentencing process.

At the outset, candor compels recognition that our cases regarding the delegation by Congress of lawmaking power do not always say what they seem to mean. Kenneth Culp Davis has been instrumental in pointing out the 'unreality' of judicial language appearing to direct attention solely to the presence or absence of statutory 'standards' or an 'intelligible principle' by which delegated authority may be guided. See generally 1 K. Davis, Administrative Law Treatise §§ 2.01 to 2.05 (1958). In his words,

'The difficulty and complexity of some types of policy     determination requires that the legislative body should be      allowed to provide for the administrative working out of      basic policy through the use of specialized tribunals which      use the common-law method of concentrating upon one      particular, narrow, and concrete problem at a time. The     protection of advance legislative guidance is of little or no      consequence as compared with the protection that can and      should be provided through adequate procedural safeguards,      appropriate legislative supervision or reexamination, and the      accustomed scope of judicial review.

'The protection that comes from a hearing with a     determination on the record, from specific findings and      reasons, from opportunity for outside critics to compare one      case with another, from critical supervision by the      legislative authority *  *  * and from judicial review-all this      is likely to be superior to protection afforded by      definiteness of standards.' Id., §§ 2.05, at 98-99, 2.09, at      111 (1958).

The point made by Professor Davis has, I think, often been recognized by Congress. It is not surprising, then, to see that in many instances Congress has focused its attention much less upon the definition of precise statutory standards than on the creation of other means adequate to assure that policy is set in accordance with congressional desires and that individuals are treated according to uniform principles rather than administrative whim. Viewed in this light, our cases may be considered as illustrating at least three legislative techniques.

First. In a number of instances, Congress has in fact undertaken to regulate even rather complex questions by the prescription of relatively specific standards. It is certainly an open question whether determining what conduct should be subject to criminal sanctions is any more difficult than determining what those sanctions should be; yet Congress and the state legislatures as well have regularly passed criminal codes embodying, in the main, statutes directed at specifically and narrowly defined conduct. Similarly, the Congress resolved what was certainly one of the most delicate and complex questions before it in recent years-the extent, if any, to which the national interest warranted federal regulation of organizations, including political parties, infiltrated by, dominated by, or subject to foreign control-not by leaving the matter to anyone else but by defining with careful particularity the characteristics that were required before an organization could be subject to such regulation. See 50 U.S.C. §§ 782(3), (4), (4A), (5) (1964 ed., Supp. V); Communist Party of United States v. SACB, 367 U.S. 1, 81 S.Ct. 1357, 6 L.Ed.2d 625 (1961). Congressional response to the complex and intractable problems of the depression are occasionally took a similar form. Thus the Act approved in United States v. Rock Royal Co-op., 307 U.S. 533, 59 S.Ct. 993, 83 L.Ed. 1446 (1939), stated a congressional policy to restore parity prices in milk, defined the term, and delegated to the Secretary of Agriculture only the power to issue orders in terms themselves specified in the Act, commanding minimum prices to be determined in accordance with prescribed standards, to be applicable in areas where prices had fallen below the limit set by Congress. See id., at 575-577, 59 S.Ct., at 1013-1015.

Second. In other circumstances, Congress has granted to others the power to prescribe fixed rules to govern future activity and adjudications. Such delegations of power permit the legislature to declare the end sought and leave technical matters in the hands of experts, or to leave to others the task of devising specific rules to carry out congressional policy in a variety of factual situations. Where, as is often the case, even major policy decisions may turn on specialized knowledge and expertise beyond legislative ken, delegation of rulemaking power may be made under broad standards to a body chosen for familiarity with the subject matter to be regulated. But entirely aside from whatever procedural protections may be afforded interested parties prior to the promulgation of administrative rules, the very nature of the rulemaking process provides significant guarantees both of evenhanded treatment and of ultimate legislative supervision of fundamental policy questions. Significantly, we have upheld delegations of rulemaking power without standards to guide its exercise only in two narrowly limited classes of cases. We have otherwise searched the statute, the legislative history, and the context in which the regulation was enacted in order to discern and articulate a legislative policy. The point is not whether an intelligible legislative policy was or was not correctly inferred from the statute. The point is that such a policy, once expressly articulated, not only serves to guide subsequent administrative and judicial action but also provides a basis upon which the legislature may determine whether power is being exercised in accordance with its will. Where no intelligible resolution of fundamental policy questions can be discerned from a statute or judicial decisions, the rulemaking process itself serves to make explicit the agency's resolution of these questions, thus allowing for meaningful legislative supervision, as well as providing bases both for judicial review of agency action supposedly premised on the rule and for refinement of an old rule in light of experience gained in its administration.

Third. Perhaps the most common legislative technique for dealing with complex questions that will arise in a myriad of factual contexts has been the delegation to another group of lawmaking power which may be exercised either through rulemaking or the adjudication of individual cases, with choice between the two left to the agency's judgment. Such schemes, while allowing broad flexibility for the working out of policy on a case-by-case basis, nevertheless have invariably provided substantial protections to insure against arbitrary action and to guarantee that underlying questions of policy are considered and resolved. As with the delegation simply of rulemaking power, we have often found substantial guidance in the language and history of the governing statute. New York Central Securities Corp. v. United States, 287 U.S. 12, 53 S.Ct. 45, 77 L.Ed. 138 (1932); Federal Radio Commission v. Nelson Bros. Co., 289 U.S. 266, 53 S.Ct. 627, 77 L.Ed. 1166 (1933); Sunshine Anthracite Coal Co. v. Adkins, 310 U.S. 381, 60 S.Ct. 907, 84 L.Ed. 1263 (1940). Agency action under such delegations must typically be premised upon an explanation of both the findings and reasons for a given decision, e.g., 5 U.S.C. § 557(c)(3) (1964 ed., Supp. V), a requirement we have held to be far more than an empty formality. SEC v. Chenery Corp., 318 U.S. 80, 63 S.Ct. 454, 87 L.Ed. 626 (1943); Phelps Dodge Corp. v. NLRB, 313 U.S. 177, 196-197, 61 S.Ct. 845, 853-854, 85 L.Ed. 1271 (1941). The regular course of adjudication by a continuing body required to explain the reasoning upon which its decisions are based results in the accumulation of a body of precedent from which, over time, general principles may be deduced. See, e.g., the history of the Federal Communications Commission's 'fairness doctrine,' traced in Red Lion Broadcasting Co. v. FCC, 395 U.S. 367, 375-379, 89 S.Ct. 1794, 1798-1801, 23 L.Ed.2d 371 (1969). We have often noted the importance of administrative or judicial review in providing a check on the exercise of arbitrary power, Mulford v. Smith, 307 U.S. 38, 49, 59 S.Ct. 648, 652-653, 83 L.Ed. 1092 (1939); American Power & Light Co. v. SEC, 329 U.S. 90, 105, 67 S.Ct. 133, 142, 91 L.Ed. 103 (1946), and we have made clear that judicial review is designed to reinforce internal protections against arbitrary or unconsidered action while leaving questions of policy to the agency or the Congress. Thus we have withheld approval from agency action unsupported by an indication of the reasons for that action, Phelps Dodge Corp. v. NLRB, supra; where the reasons articulated were improper, Sicurella v. United States, 348 U.S. 385, 75 S.Ct. 403, 99 L.Ed. 436 (1955), even though the record might well support identical action taken for different reasons, SEC v. Chenery Corp., supra; where administrative expertise relevant to the solution of a problem had never been brought to bear upon it, FCC v. RCA Communications, Inc., 346 U.S. 86, 91-92, 73 S.Ct. 998, 1002-1003, 97 L.Ed. 1470 (1953); where an apparent conflict in administrative rationales had never been explained by the agency, Barrett Line, Inc. v. United States, 326 U.S. 179, 65 S.Ct. 1504, 89 L.Ed. 2128 (1945); and where a change in agency policy had taken place after the particular adjudication concerned, NLRB v. Gissel Packing Co., 395 U.S. 575, 615-616, 89 S.Ct. 1918, 1940-1941, 23 L.Ed.2d 547 (1969).

Combination of rulemaking and adjudicatory powers has proved a particularly useful tool in situations where prescription of detailed standards in the first instance has been difficult or impossible for the Congress, yet the variety of factual situations has rendered particularly important protection against random or arbitrary decisions. Thus in Lichter v. United States, 334 U.S. 742, 68 S.Ct. 1294, 92 L.Ed. 1694 (1948), this Court dealt with the provisions of the original Renegotiation Act, passed in April 1942, which directed various administrative officials to proceed with compulsory 'renegotiation' of contracts that had resulted in 'excessive profits.' The Act as originally passed attempted no definition of such profits; within four months, however, administrative practice had solidified about a list of six factors to be considered in determining whether profits were excessive; slightly more than two months later, these factors were adopted by Congress in an amendment to the Act. In upholding the original Act against a claim of excessive delegation, we stressed both the rapid development of generally applicable standards, id., at 766, 769, 771, 773-774, 778, 783, 68 S.Ct., at 1307, 1309, 1310, 1311, 1313, 1315, and the availability of judicial review to check arbitrary or inconsistent administrative action. Id., at 770, 771, 786-787, 68 S.Ct., at 1309, 1310, 1317-1318.

The next question is whether there is anything inherent in the nature of capital sentencing that makes impossible the application of any or all of the means that have been elsewhere devised to check arbitrary action. I think it is fair to say that the Court has provided no explanation for its conclusion that capital sentencing is inherently incapable of rational treatment. Instead, it relies primarily on the Report of the (British) Royal Commission on Capital Punishment, which reaches conclusions substantially identical with the following urged in 1785 by Archdeacon William Paley to justify England's 'Bloody Code' of more than 250 capital crimes:

'(T)he selection of proper objects for capital punishment     principally depends upon circumstances, which, however easy      to perceive in each particular case after the crime is      committed, it is impossible to enumerate or define      beforehand; or to ascertain, however, with that exactness,      which is requisite in legal descriptions. Hence, although it     be necessary to fix, by precise rules of law, the boundary on      one side *  *  * yet the mitigation of punishment *  *  * may,      without danger, be intrusted to the executive magistrate,      whose discretion will operate upon those numerous,      unforeseen, mutable and indefinite circumstances, both of the      crime and the criminal, which constitute or qualify the      malignity of each offence. * *  * For if judgment of death      were reserved for one or two species of crimes only *  *  *      crimes might occur of the most dangerous example, and      accompanied with circumstances of heinous aggravation, which      did not fall within any description of offences that the laws      had made capital, and which, consequently, could not receive      the punishment their own malignity and the public safety      required. * *  *

'The law of England is constructed upon a different and a     better policy. By the number of statutes creating capital     offences, it sweeps into the net every crime, which under any      possible circumstances may merit the punishment of death:      but, when the execution of this sentence comes to be      deliberated upon, a small proportion of each class are      singled out, the general character, or the peculiar      aggravations of whose crimes, render them fit examples of      public justice. * *  * The wisdom and humanity of this design furnish a just excuse for the multiplicity of capital      offences, which the laws of England are accused of creating      beyond those of other countries.' W. Paley, Principles of      Moral and Political Philosophy 399-401 (6th Amer. ed. 1810).

Significantly, the Court neglects to mention that the recommendations of the Royal Commission on Capital Punishment found little more favor in England than Archdeacon Paley's. For the 'British have been unwilling to empower either courts or juries to decide on life or death, insisting that death should be the sentence of the law and not of the tribunal.' Symposium on Capital Punishment, 7 N.Y.L.F. 249, 253 (1961) (H. Wechsler). Beyond the Royal Commission's Report, the Court supports its conclusions only by referring to the standards proposed in the Model Penal Code and judging them less than perfect. The Court neglects to explain why the impossibility of perfect standards justifies making no attempt whatsoever to control lawless action. In this context the words of Mr. Justice Frankfurter are instructive:

'It is not for this Court to formulate with particularity the     (standards) which would satisfy the Fourteenth Amendment. No     doubt, finding a want of such standards presupposes some      conception of what is necessary to meet the constitutional      requirement we draw from the Fourteenth Amendment. But many a     decision of this Court rests on some inarticulate major      premise and is none the worse for it. A standard may be found     inadequate without the necessity of explicit delineation of      the standards that would be adequate, just as doggerel may be      felt not to be poetry without the need of writing an essay on what poetry is.' Niemotko v. Maryland, 340 U.S., at 285,      71 S.Ct., at 334 (concurring in result).

But, although I find the Court's discussion inadequate, there remains the question whether capital sentencing is inherently incapable of being carried out under procedures that provide the safeguards necessary to protect against arbitrary determinations. I think not. I reach this conclusion for the following reasons.

First. It is important at the outset to recognize that two separate questions are involved. The first question is what ends any given State seeks to achieve by imposing the death penalty. The second question is whether those ends will or will not be served in any given case. The first question requires determination of the penological policy adopted by the State in choosing to kill some of its convicted criminals. The second question requires that the relevant facts in any particular case be determined, and that the State's penological policy be applied to those facts.

Second. It is likewise important to bear in mind that the complexity of capital sentencing in any particular jurisdiction is inevitably a function of the penological policy to be applied. It is not, inherently, a difficult question. Thus, if a State should determine to kill those first-degree murderers who have been previously convicted of murder, and only those persons, the sentencing determination would ordinarily be a rather simple one. On the other hand, if a State should determine to exclude only those first-degree murderers who cannot be rehabilitated, it is probably safe to assume that the question of proper sentencing under such a policy would be a complex one indeed. It should be borne in mind that either of these policies-or a host of others-may have been applied in the cases before us.

Third. This is neither the time nor the place for an essay on the purposes of criminal punishment. Yet some discussion must be ventured. Without indicating any judgment as to their propriety and without intending to suggest that no others may exist-it is apposite to note that the interests most often discussed in connection with a State's capital sentencing policy are four. A State may seek to inflict retribution on a wrongdoer, inflicting punishment strictly in proportion to the offense committed. It may seek, by the infliction of punishment, to deter others from committing similar crimes. It may consider at least some wrongdoers likely to commit other crimes, and therefore seek to prevent these hypothetical future acts by removing such persons from society. It may seek to rehabilitate most offenders, reserving capital punishment only for those cases where it judges the likelihood of rehabilitation to be less than a certain amount. I may assume that many if not all States choosing to kill some convicted criminals intend thereby to further more than one of the ends listed above; and I need not doubt that some States may consider other policies as well relevant to the decision. But I can see no reason whatsoever that a State may be excused from declaring what policies it seeks to further by the infliction of capital punishment merely because it may be difficult to determine how those policies should be applied in any particular case. If anything, it would seem that the difficulty of decision in particular cases would support rather than weaken the point that uniform decisionmaking requires that state policy be explicitly articulated. Yet the Court seems somehow to assume that jurors will be most likely to fulfill their function and correctly apply a uniform state policy if they are never told what that policy is. If this assumption finds support anywhere this side of the Looking-Glass World, I am unaware of it.

Fourth. This is not to say, of course, that there may be no room whatsoever for the exercise of discretion in the capital sentencing process. But discretion, to be worthy of the name, is not unchanneled judgment; it is judgment guided by reason and kept within bounds. Otherwise, in Lord Camden's words, it is 'the law of tyrants: It is always unknown: It is different in different men: It is casual, and depends upon constitution, temper, passion. In the best it is oftentimes caprice: In the worst it is every vice, folly, and passion, to which human nature is liable.' Hindson and Kersey, cited in 8 How.St.Tr. 57 n. It may well be that any given State's notions of proper penological policy are such that the precise amount of weight to be given to any one factor in any particular case where death is a possible penalty is incapable of determination beforehand. But that is no excuse for refusing to tell the decisionmaker whether he should consider a particular factor at all. Particularly where decisions are made, not by a continuing body of persons, but by groups selected to make a single decision and dispersed immediately after the event, the likelihood of any consistency whatsoever is vanishingly small. 'Perfection may not be demanded of law, but the capacity to counteract inevitable * *  * frailties is the mark of a civilized legal mechanism.' Rosenberg v. United States, 346 U.S. 273, 310, 73 S.Ct. 1152, 1171, 97 L.Ed. 1607 (1953) (Frankfurter, J., dissenting). The point is that even if a State's notion of wise capital sentencing policy is such that the policy cannot be implemented through a formula capable of mechanical application-something which, incidentally, cannot be known unless and until the State makes explicit precisely what that policy is-there is no reason that it should not give some guidance to those called upon to render decision.

Fifth. As I have already indicated, typical legislative response to problems deemed of sufficient urgency that some solution must be implemented immediately, yet at the same time of sufficient difficulty as to be incapable of explicit statutory solution, has been to provide a means whereby the law may be usefully developed on a case-by-case basis: systems are devised whereby each case may be decided upon its facts, with consistency and the development of more general principles left to the wisdom that comes from experience. I am speaking, of course, of the administrative process, where the basis and reasons for any given decision are explained and subject to review. I see no reason that capital sentencing is ipso facto unsuited to such treatment. To begin with, if a legislature should deem its present knowledge insufficient to create proper standards, it is hard indeed to see why its solution should not be one that could ultimately lead to the development of such standards. Cf. Lichter v. United States, 334 U.S. 742, 68 S.Ct. 1294, 92 L.Ed. 1694 (1948). I see no reason that juries which have determined that a given person should be killed by the State should be unable to explain why they reached that decision, and the facts upon which it was based. Persons dubious about the ability of juries to explain their findings should consult Fletcher v. Peck, 10 U.S. (6 Cranch) 87, 95-114, 3 L.Ed. 162 (1810) (findings of trial jury). Cf. Fed.Rule Civ.Proc. 49. Even if it be assumed that juries are incapable of making such explanations, we have already held that such inability does not excuse the State from providing a sentencing process that provides reasons for the decisions reached if those reasons are otherwise required. North Carolina v. Pearce, 395 U.S. 711, 726, 89 S.Ct. 2072, 2081, 23 L.Ed.2d 656 (1969).

In sum, I see no reason whatsoever to believe that the nature of capital sentencing is such that it cannot be surrounded with the protections ordinarily available to check arbitrary and lawless action. That it has not been is, of course, no reason to believe that it cannot be:

'As to impossibility, all I can say is that nothing is more     true of (the legal) profession than that the most eminent      among them, for 100 years, have testified with complete      confidence that something is impossible which, once it is      introduced, is found to be very easy of administration. The     history of legal procedure is the history of rejection of      reasonable and civilised standards in the administration of      law by most eminent judges and leading practitioners. * *  *      Every effort to effect improving changes is resisted on the      assumption that man's ultimate wisdom is to be found in the      legal system as at the date at which you try to make a      change.' F. Frankfurther, The Problem of Capital Punishment,      in Of Law and Men 77, 86 (1956).

I have explained above the reasons for my belief that the Due Process Clause of the Fourteenth Amendment compels the States to make explicit the fundamental policy choices upon which any exertion of state power is based, and to exercise such power only under procedures that both limit the possibility of merely arbitrary action and provide a record adequate to render meaningful the institution of federal judicial review. I have also explained why, in my view, there is nothing inherent in the nature of capital sentencing that makes application of such procedures impossible. There remains, then, only the question whether the two state procedures under review today provide the necessary safeguards.

* In Ohio, if a capital defendant elects trial by jury the questions whether he is guilty of the crime charged and, if so, whether he should be killed are simultaneously submitted to the jury. Jury trial may, however, be waived as of right in capital cases, State v. Smith, 123 Ohio St. 237, 174 N.E. 768 (1931), or a defendant may, with the permission of the court, enter a plea of guilty. State v. Ferranto, 112 Ohio St. 667, 148 N.E. 362 (1925). In the absence of jury trial the sentencing decision is made by a three-judge court. Ohio Rev. Code Ann. § 2945.06 (1954).

A defendant who exercises his right to jury trial may introduce only evidence relevant to the question of guilt. No evidence may 'be introduced directed specifically toward a claim for mercy,' Ashbrook v. State, 49 Ohio App. 298, 302, 197 N.E. 214, 216 (1935), for that 'is a matter vested fully and exclusively in the discretion of the jury,' State v. Ellis, 98 Ohio St. 21, 120 N.E. 218 (court's syllabus) (1918), and therefore, under Ohio law, is 'not an issue in the case.' Ashbrook v. State, supra. A defendant who can present no evidence on the question of guilt may not, therefore, present any evidence whatsoever to the sentencing jury.

A defendant who waives jury trial, however, is in a somewhat different situation. Presumably, of course, the same rules of evidence apply at a bench trial or at a trial upon a plea of guilty. Where the sentencing determination is made by the court, however, two additional factors apply. First, the defendant has an absolute right to address the court before sentence is imposed, Ohio Rev. Code Ann. § 2947.05 (1954), denial of which is a ground for resentencing. Silsby v. State, 119 Ohio St. 314, 164 N.E. 232 (1928). Since the jury's decision that a defendant should be killed is unreviewable by any court, State v. Klumpp, Ohio App., 15 Ohio Op.2d 461, 468, 175 N.E.2d 767, 775-776, appeal dismissed, 171 Ohio St. 62, 167 N.E.2d 778 (1960) (trial court); State v. Reed, 85 Ohio App. 36, 84 N.E.2d 620 (1948), exercise of this right can have no effect on the sentencing determination in jury cases. But the trial court may modify its own sentence during the same term of court, see Lee v. State, 32 Ohio St. 113 (1877), and may therefore be swayed by the defendant's personal plea. Moreover, Ohio Rev. Code Ann. § 2947.06 (Supp.1970) expressly permits a trial court to 'hear testimony of mitigation of a sentence at the term of conviction or plea.' If this statute is applicable to capital cases, defendants pleading guilty or waiving jury trial may introduce additional information on the question of sentence. Again, however, the unreviewability of a jury sentence means that it can have no effect in cases tried to a jury. Finally, a death sentence imposed by a three-judge court may not be reviewed or modified on appeal. State v. Ferguson, 175 Ohio St. 390, 195 N.E.2d 794 (1964); State v. Stewart, 176 Ohio St. 156, 198 N.E.2d 439 (1964).

The standard instruction given capital juries on the question of punishment appears in State v. Caldwell, 135 Ohio St. 424, 425, 21 N.E.2d 343,344 (1939):

'(Y)ou will determine whether or not you will extend or     withhold mercy. * *  * In that connection whether you recommend or withhold mercy is a matter solely      within your discretion, calling for the exercise of your very      best and most profound judgment, not motivated by      considerations of sympathy or as a means of escaping a hard      or disagreeable duty, but must be considered by you in the      light of all the circumstances of the case with respect to      the evidence submitted to you and the other circumstances      surrounding this defendant.'

The jury may be instructed that 'sociological matters and environment' have 'nothing whatever to do with (the) case,' id., at 428, 21 N.E.2d, at 344, but it appears that this instruction is not generally given. Likewise, the trial court may, but is not compelled to, inform the jury about matters such as parole from a sentence to life imprisonment. State v. Meyer, 163 Ohio St. 279, 126 N.E.2d 585 (1955); State v. Henley, 15 Ohio St.2d 86, 238 N.E.2d 773 (1968). In petitioner Crampton's case, the jury was instructed generally that it should not be 'influenced by any consideration of sympathy or prejudice.' On the question of punishment it was told only that '(i)f you find the defendant guilty of murder in the first degree, the punishment is death, unless you recommend mercy, in which event the punishment is imprisonment in the penitentiary during life.' The jury was also handed a verdict form with a 'line which you must fill in. We blank-recommend mercy and you will put in that line, we do, or, we do not, according to your finding.' Except for a supplementary instruction informing the jury that its recommendation had to be unanimous, no further instructions on the question of punishment were given the jury.

There is in my view no way that this Ohio capital sentencing procedure can be thought to pass muster under the Due Process Clause.

First. Nothing whatsoever in the process either sets forth the basic policy considerations that Ohio believes relevant to capital sentencing, or leads towards elucidation of these considerations in the light of accumulated experience. The standard jury instruction contains at best an obscure hint. The instructions given in the present case contain none whatsoever. So far as they are concerned, the jury could have decided to impose the death penalty as a matter of simple vengeance for what it considered an atrocious crime; because it felt that imposition of the death penalty would deter other potential murderers; or because it felt that petitioner, if not himself killed, might kill or commit some other wrong in the future. The jury may have been influenced by any, all, or none of these considerations. If it is beyond the present ability of the Ohio Legislature to 'identify before the fact those characteristics of criminal homicides and their perpetrators which'-in the judgment of the State of Ohio 'call for the death penalty, and to express these characteristics in language which can be fairly understood and applied by the sentencing authority,' ante, at 204, the Ohio procedure is hardly designed to improve that ability. It contains no element of the proudest tradition of the common law-the ability to grow with time by slowly deriving principles of general applicability from careful consideration of the myriad facts of a multitude of particular cases. Neither we nor the State of Ohio can know the reasoning by which this jury determined to impose the death penalty, or the facts upon which that reasoning was based. All we know is that the jury did not appear to find the question a particularly difficult one. For the jury determined that James Edward Crampton had murdered his wife, that he had done so while legally sane, and that he should be killed-in less than five hours.

Second. The policies applied by the State of Ohio to determine that James Edward Crampton should die were neither articulated to nor explained by the jury that made that decision. Nor have they been elsewhere set forth. The standard jury instructions, quoted supra, at 289-290, do tell the jury to reach its determination 'in the light of all the circumstances of the case with respect to the evidence submitted to you and the other circumstances surrounding this defendant.' A perceptive jury might conclude that this instruction indicates that Ohio considers the relative severity of the crime a factor of substantial importance in the determination of sentence. How the jury is to determine the severity of the crime before it in relation to others is, however, something of a mystery, since Ohio law simultaneously demands that the sentencing determination be based strictly upon the evidence adduced in the case at hand, Howell v. State, 102 Ohio St. 411, 131 N.E. 706 (1921), and forbids the defendant to introduce evidence of other crimes or other judgments to aid the jury in determining whether the murder he has committed is more or less severe than other murders. Ashbrook v. State, supra. Similarly, by directing the jury's attention to 'the other circumstances surrounding this defendant' it might be thought that Ohio was suggesting consideration of environmental factors that might make the defendant's actions, if no more justifiable, less a reflection of personal blameworthiness. Yet any such reading of the instruction is condemned by State v. Caldwell, supra, which approved a jury charge that environmental factors have 'nothing whatever to do' with the sentencing decision. It also might be thought that directing juries to consider 'other circumstances surrounding this defendant' is an indication, albeit a rather backhanded one, that Ohio desires capital sentencing juries to take into account the likelihood that a particular defendant may be rehabilitated. Certainly this indication is reinforced in cases where the jury is instructed with regard to the possibility of parole from a life sentence. But instructions on parole are optional with the trial court, State v. Henley, supra; State v. Meyer, supra, and unless it be assumed that every jury not so instructed is nevertheless aware of the possibility of parole (and likewise that, despite instructions to base its verdict on the evidence in the case, it will nevertheless rely upon its own knowledge of the possibility of parole), failure to instruct all juries with regard to parole must mean either that a state policy with regard to rehabilitation is not in fact implied by such instructions, or else that such a state policy is consciously applied only in some capital cases. Finally, one Ohio case may be explicable only on a basis suggested nowhere else in Ohio law: that the capital sentencing decision rests upon factors that vary depending upon which of two simultaneously applicable capital statutes is used to support punishment. In State v. Ferguson, 175 Ohio St. 390, 195 N.E.2d 794 (1964), the defendant had been convicted on guilty pleas entered to charges of premeditated murder and felony murder, both growing out of the murder, during the course of a robbery, of a single individual. The three-judge court that heard evidence to fix the penalty on both charges at the same time sentenced him to life imprisonment on the premeditated murder charge, and to death on the charge of felony murder. The Ohio Supreme Court affirmed the sentence of death. In light of these cases, I think it fair to say that Ohio law has nowhere purported to set forth the considerations of state policy intended to underlie a sentence of death.

Third. Even if it be assumed that Ohio sentencing judges and juries act upon shared, although unarticulated and unarticulable, notions of proper capital sentencing policy, the capital sentencing process in Ohio contains elements that render difficult if not impossible any consistency in result. Presumably all judges, and certainly some juries (i.e., those who are specifically so instructed) will be cognizant of the possibility of parole from a sentence to life imprisonment. Other juries will not. If this is an irrelevant factor, it is hard to understand why some juries may be given this information. If it is a relevant factor, it is equally hard to understand why other juries are not. And if it is a relevant factor, the inevitable consequence of presenting the information, for no explicable reason, to some but not all capital sentencing juries, is that consistency in decisionmaking is impossible. Similarly, as I have already noted, there is a substantial difference between the evidence that may be considered by a jury and that which may be considered by a sentencing panel of judges. For although the defendant may, in a jury trial, testify on the question of guilt if he is willing to forgo his privilege against self-incrimination, he may not even then present evidence relevant solely to the question of penalty. A defendant who is to be sentenced by a panel of judges, on the other hand, has an absolute right before the sentencing decision becomes final to address the sentencers on any subject he may choose. And such a defendant appears as well to have at least a chance to present evidence from other sources relevant solely to the sentencing determination before that determination becomes final. Yet such information may not be presented to a jury, whether the jury desires it or not. The point, again, is that consistent decisionmaking is impossible when one decisionmaker may consider information forbidden to another. And where, as here, no basis whatsoever is presented to justify the difference, it is inexcusable.

Fourth. There is, moreover, no reason to believe that Ohio capital sentencing judges and juries do in fact share common notions of the considerations relevant to capital sentencing. I have already pointed out that no state policy has ever been articulated. And whatever may be the case with judges, capital sentencing juries are drawn essentially at random and called upon to decide one case and one case only. Whatever value there may be in the notion that arbitrary decisionmaking may be controlled by committing difficult questions to a continuing body which can at least maintain consistency of principle until it changes its views on the questions to be decided, is entirely absent from the capital jury sentencing process presently under review. For capital sentencing juries in Ohio are not continuing bodies, and no jury may be told what another jury has done in similar (or different) cases. Likewise, the procedure under review cannot gain uniformity from judicial review, for under Ohio law no such review is permitted.

Fifth. Although the Due Process Clause does not forbid a State from imposing 'a different punishment for the same offence * *  * under particular circumstances,' Moore v. Missouri, 159 U.S. 673, 678, 16 S.Ct. 179, 181, 40 L.Ed. 301 (1895), it does command that punishment be 'dealt out to all alike who are similarly situated.' Ibid.; Leeper v. Texas, 139 U.S. 462, 468, 11 S.Ct. 577, 579, 35 L.Ed. 225 (1891); Missouri v. Lewis, 101 U.S. 22, 31, 25 L.Ed. 989 (1880). Even granting the State the fullest conceivable room for judgment as to what are and are not 'particular circumstances' justifying different treatment, this means at the least that the State must itself apply the same fundamental policies to all in making that judgment. The institution of federal judicial review is designed to vindicate this (and other ) federally guaranteed rights. Yet the procedure before us renders the possibility of such review entirely chimerical. There is no way of determining what policies were applied by the State in reaching judgment. There is no way of inferring what policies were applied by an examination of the facts, for we have no idea what facts were relied on by the sentencers. Nor may this void be filled in any way by presumptions based on the result of their actions, for they were neither given direction in the exercise of judgment nor asked to explain the conclusion they reached. There is simply no way that this or any other court can determine whether petitioner Crampton was condemned to die for reasons that Ohio would be willing to apply in any other case-or for reasons that Ohio would, if they were explicitly set forth, just as explicitly reject.

In sum, the Ohio capital sentencing procedure presently before us raises fundamental questions of state policy which have never been explicitly decided by any responsible organ of the State. Nothing in the procedure looks towards the gradual development of a uniform state policy through accumulation of a body of precedent. No protection whatsoever appears against the possibility of merely arbitrary or willful decisionmaking; moreover, some features of the process appear to make inconsistent action not merely possible but inevitable. And finally, the record provided by the Ohio capital sentencing process makes virtually impossible the redress of any violations of federally guaranteed rights through the institution of federal judicial review. I can see no pssible basis for holding such a capital sentencing procedure permissible under the Due Process Clause, and I would therefore reverse petitioner Crampton's sentence of death.

The procedures whereby the State of California determines which convicted criminals to kill differ in a number of respects from those used by Ohio. Following conviction of a possibly capital crime, the question of penalty is determined in a separate proceeding. Except where the defendant has, with the prosecution's consent, waived trial by jury, the sentencing determination is made by a jury whether conviction was on a plea of guilty or not guilty. A defendant who waives jury trial on the issue of guilt may not have his sentence determined by a jury. People v. Golston, 58 Cal.2d 535, 25 Cal.Rptr. 83, 375 P.2d 51 (1962). Notwithstanding the statutory language, it appears possible for a defendant whose guilt is determined by a jury to have his sentence determined by a judge. See People v. Sosa, 251 Cal.App.2d 9, 58 Cal.Rptr. 912 (1967). If a jury is waived, identical sentencing power will be exercised by a single judge. People v. Langdon, 52 Cal.2d 425, 341 P.2d 303 (1959); People v. Jones, 52 Cal.2d 636, 343 P.2d 577 (1959). A jury determination to impose a death sentence may be set aside by the judge presiding at the trial, Cal.Penal Code § 1181(7) (1970), construed in People v. Hill, 66 Cal.2d 536, 58 Cal.Rptr. 340, 426 P.2d 908 (1967). It may not be otherwise reviewed, whether fixed by a judge or jury. People v. Welch, 58 Cal.2d 271, 23 Cal.Rptr. 363, 373 P.2d 427 (1962) (judge); In re Anderson, 69 Cal.2d 613, 73 Cal.Rptr. 21, 447 P.2d 117 (1968).

The range of evidence that may be introduced at the penalty trial is broad. Ordinary rules of competence, hearsay, etc., apply, e.g., People v. Hines, 61 Cal.2d 164, 174-175, 37 Cal.Rptr. 622, 629, 390 P.2d 398, 405 (1964), and a few issues are excluded. Exclusion, however, appears to be not on the basis that the issues are irrelevant, but rather that they are either unduly inflammatory or impractical to litigate. Thus, evidence or argument is prohibited concerning the likelihood of parole from a life sentence, People v. Morse, 60 Cal.2d 631, 36 Cal.Rptr. 201, 388 P.2d 33 (1964); concerning the deterrent effects of capital punishment, People v. Purvis, 60 Cal.2d 323, 341, 33 Cal.Rptr. 104, 116, 384 P.2d 424, 435-436 (1963); People v. Love, 56 Cal.2d 720, 16 Cal.Rptr. 777, 366 P.2d 33 (1961); People v. Kidd, 56 Cal.2d 759, 16 Cal.Rptr. 793, 366 P.2d 49 (1961), although some reference to the matter may (as in the present case, see App. 199) be made by the prosecution and be treated under the harmless-error doctrine, People v. Garner, 57 Cal.2d 135, 18 Cal.Rptr. 40, 367 P.2d 680 (1961), especially if trial is to the court, People v. Welch, 58 Cal.2d 271, 274, 23 Cal.Rptr. 363, 365, 373 P.2d 427, 429 (1962); concerning whether capital punishment should ever be imposed, People v. Moya, 53 Cal.2d 819, 3 Cal.Rptr. 360, 350 P.2d 112 (1960); or concerning physical suffering of the victim unintended by the defendant, People v. Love, 53 Cal.2d 843, 3 Cal.Rptr. 665, 350 P.2d 705 (1960). Except for these limitations, however, virtually any matter may be explored. People v. Terry, 61 Cal.2d 137, 142-153, 37 Cal.Rptr. 605, 609-619, 390 P.2d 381, 385-392 (1964).

Following the arguments of counsel, the jury is instructed on its function in determining the penalty to be imposed. A standard instruction on the subject exists but is not mandatory; it is, essentially, the instruction given in the present case:

'The defendants in this case have been found guilty of the     offense of murder in the first degree, and it is now your      duty to determine which of the penalties provided by law      should be imposed on each defendant for that offense. Now, in     arriving at this determination you should consider all of the      evidence received here in court presented by the People and      defendants throughout the trial before this jury. You may     also consider all of the evidence of the circumstances      surrounding the crime, of each defendant's background and      history, and of the facts in aggravation or mitigation of the      penalty which have been received here in court. However, it     is not essential to your decision that you find mitigating      circumstances on the one hand or evidence in aggravation of      the offense on the other hand.

'It is the law of this state that every person guilty of     murder in the first degree shall suffer death or confinement      in the state prison for life, at the discretion of the jury. If you should fix the penalty as confinement for life, you     will so indicate in your verdict. If you should fix the     penalty as death, you will so indicate in your verdict. Notwithstanding facts, if any, proved in mitigation or     aggravation, in determining which punishment shall be      inflicted, you are entirely free to act according to your own      judgment, conscience, and absolute discretion. That verdict     must express the individual opinion of each juror.

'Now, beyond prescribing the two alternative penalties, the     law itself provides no standard for the guidence of the jury in the selection of the penalty, but,      rather, commits the whole matter of determining which of the      two penalties shall be fixed to the judgment, conscience, and      absolute discretion of the jury. In the determination of that     matter, if the jury does agree, it must be unanimous as to      which of the two penalties is imposed.'

Substantially more elaborate versions of this instruction may, if the trial court desires, be given. People v. Harrison, 59 Cal.2d 622, 30 Cal.Rptr. 841, 381 P.2d 665 (1963). In addition, the trial court is supposed to instruct the jury that a defendant serving a life sentence may be paroled, but that it should not presume that the California Adult Authority will release a prisoner until it is safe to do so, and that it should not take the possibility of parole into account. People v. McGautha, 70 Cal.2d 770, 76 Cal.Rptr. 434, 452 P.2d 650 (1969). Finally, under California law it is error to charge that the jury's verdict should express the conscience of the community; the jury should be told, instead, that the verdict must 'express the individual conscience of each juror.' People v. Harrison, supra, 59 Cal.2d, at 633, 30 Cal.Rptr., at 847, 381 P.2d, at 671.

A substantial number of subsidiary instructions may but need not be given to the jury; the governing principle is that the instructions must make clear to the jury that its decision whether or not a convicted defendant is to be killed is to take place in a 'legal vacuum.' People v. Terry, 61 Cal.2d, at 154, 37 Cal.Rptr. at 616, 390 P.2d, at 392; see People v. Friend, 47 Cal.2d 749, 306 P.2d 463 (1957). A trial judge may, should he desire, 'aid the jury by stating the kinds of factors that may be considered, thereby setting the tone for the jury's deliberations,' People v. Polk, 63 Cal.2d 443, 451, 47 Cal.Rptr. 1, 6, 406 P.2d 641, 646 (1965), so long as this is done in a manner that indicates to the jury that it is free not to consider any of the factors listed by the judge, and to consider anything else it may desire, People v. Friend, supra. It is not, however, error to refuse such an instruction. People v. Polk, supra. Similarly, although a trial judge may instruct the jury that it may be moved by sympathy for the defendant, People v. Anderson, 64 Cal.2d 633, 51 Cal.Rptr. 238, 414 P.2d 366 (1966), he may refuse to give such an instruction at defense request, People v. Hillery, 65 Cal.2d 795, 56 Cal.Rptr. 280, 423 P.2d 208 (1967), although it is error to instruct the jury that it may not be so moved. People v. Polk, supra; People v. Bandhauer, 1 Cal.3d 609, 83 Cal.Rptr. 184, 463 P.2d 408 (1970). It is error to instruct the jury that it may not consider doubts about the defendant's guilt as mitigating circumstances, People v. Terry, supra, but it is not error to refuse to charge that such doubt may be a mitigating factor, People v. Washington, 71 Cal.2d 1061, 80 Cal.Rptr. 567, 458 P.2d 479 (1969), although the trial judge may give such a charge if he desires, People v. Polk, supra; People v. Terry, supra.

Finally, a jury determination to impose the death sentence may not be reviewed by any court. It may, however, be set aside by the judge presiding at the trial. The basis upon which the California Supreme Court has made this distinction, of some importance in the present case, is not entirely clear. The trial judge's power to reduce a sentence of death to one of life imprisonment is based on Cal.Penal Code § 1181(7) (1970), which provides, in pertinent part, that 'in any case wherein authority is vested by statute in the trial court or jury to recommend or determine as a part of its verdict or finding the punishment to be imposed, the court may modify such verdict or finding by imposing the lesser punishment without granting or ordering a new trial, and this power shall extend to any court to which the case may be appealed.' The California Supreme Court has construed this statute to empower the trial court to set aside a jury verdict of death, People v. Moore, 53 Cal.2d 451, 454, 2 Cal.Rptr. 6, 8, 348 P.2d 584, 586 (1960), but not to give any such power to an appellate court, People v. Green, 47 Cal.2d 209, 235, 302 P.2d 307, 324-325 (1956); In re Anderson, 69 Cal.2d 613, 73 Cal.Rptr. 21, 447 P.2d 117 (1968). This is said to be because 'the trier of fact is vested with exclusive discretion to determine punishment.' People v. Green, supra, 47 Cal.2d, at 235, 302 P.2d, at 325. What this means is that the trial court does not review the jury's determination that a convicted defendant should be killed: based upon its 'own independent view of the evidence,' People v. Love, 56 Cal.2d, at 728, 16 Cal.Rptr., at 780, 366 P.2d, at 36, quoting People v. Moore, supra, 53 Cal.2d, at 454, 2 Cal.Rptr., at 8, 348 P.2d, at 586, the trial court is to determine itself whether the defendant should be killed, apparently on exactly the same basis and in exactly the same way as it would if the issue had never been submitted to a jury. See People v. Moore, supra; People v. Hill, 66 Cal.2d 536, 58 Cal.Rptr. 340, 426 P.2d 908 (1967); People v. Love, supra; In re Anderson, supra. In short, no defendant sentenced to die may obtain judicial review of that decision, but one sentenced to die by a jury gets a second bite at the apple: he is 'entitled to two decisions on the evidence.' People v. Ketchel, 59 Cal.2d 503, 546, 30 Cal.Rptr. 538, 561, 381 P.2d 394, 417 (1963).

I find this procedure likewise defective under the Due Process Clause. Although it differs in some not insignificant respects from the procedure used in Ohio, it nevertheless is entirely bare of the fundamental safeguards required by due process.

First. Both procedures contain at their heart the same basic vice. Like Ohio, California fails to provide any means whereby the fundamental questions of state policy with regard to capital sentencing may be authoritatively resolved. They have not been resolved by the state legislature, which has committed the matter entirely to whatevery judge or jury may exercise sentencing authority in any particular case. But they cannot be authoritatively resolved by the sentencing authority, not only because the California Supreme Court has expressly ruled that that is not part of the sentencing function, People v. Kidd, 56 Cal.2d, at 770, 16 Cal.Rptr., at 800, 366 P.2d, at 56, but also because any such resolution is binding for one case and one case only. There are simply no means to assure that 'truly fundamental issue (will ultimately) be resolved by the Legislature,' Wilke & Holzheiser, Inc. v. Department of Alcoholic Beverage Control, 65 Cal.2d 349, 369, 55 Cal.Rptr. 23, 36, 420 P.2d 735, 748 (1965). Nothing whatsoever anywhere in the process gives any assurance that one defendant will be sentenced upon notions of California penological policy even vaguely resembling those applied to the next.

Second. If the question before us were what procedure would produce the fewest number of death sentences, the power of a trial judge to set aside a jury's verdict might be of substantial importance. But that, of course, is not the question. Except insofar as it incorporates the Eighth Amendment's prohibition against cruel and unusual punishments-not an issue in these cases-the Due Process Clause gives no warrant to interfere with a State's decision to make certain crimes punishable by death. The Due Process Clause commands us, however, to make certain that no State takes one man's life for reasons that it would not apply to another. And even if it be assumed that trial judges obey the California Supreme Court's direction to exercise ther own, independent judgment on the propriety of a jury-imposed death penalty, the existence of the trial court's power to set aside such verdicts adds little to the likelihood of evenhanded treatment. For this power is to be exercised in precisely the same way as the jury's-without guideline or check, without review, without any explanation of reasons or findings of fact, without any opportunity for ultimate legislative acceptance or rejection of the policies applied. It is true that trial judges are in a sense 'professional sentencers'; presumably any given judge, to the extent that he actually does exercise independent judgment on the question, will do his best to avoid conscious inconsistency. But there remains a multiplicity of sentencing judges, all of whom have been expressly told by the Supreme Court of California not to seek guidance for their decision from the statute, from that court's opinions, or indeed from any source outside their own, individual opinions. See supra, at 304-305. In such circumstances, the possibility of consistent decision-making is nonexistent. 'A multiplicity of tribunals and a diversity of procedures are quite as apt to produce incompatible or conflicting adjudications as are different rules of substantive law.' Garner v. Teamsters Local 776, 346 U.S. 485, 490-491, 74 S.Ct. 161, 166, 98 L.Ed. 228 (1953).

Third. Like its Ohio counterpart, the California procedure before us inevitably operates to frustrate the institution of federal judicial review. We do not and cannot know what facts the jury relied upon in determining that petitioner McGautha should be killed, or the reasons upon which it based that decision. We do not know-and cannot know-the basis upon which the State of California determined that he was not 'fit to live,' People v. Morse, 60 Cal.2d, at 647, 36 Cal.Rptr., at 211, 388 P.2d, at 43. We do know that the prosecutor, in her closing argument, strongly urged to the jury that Dennis Councle McGautha should be killed because he had the unregenerate bad taste to insist that he had once pleaded guilty to a crime he did not commit. Cf. North Carolina v. Alford, 400 U.S., at 32-39, 91 S.Ct., at 165-168, 27 L.Ed.2d 162. We also know that nothing in the instructions given the jury contained the slightest hint that this could not be the sole basis for its decision. See supra, at 301-302. And, finally, we also know that whatever factors the State of California relied upon to sentence petitioner McGautha to death-factors permissible or impermissible, applied by the State to every convicted capital criminal or to him alone-there is no way whatsoever that petitioner can demonstrate that those factors were relied upon and obtain review of their propriety. In short, the procedure before us in this case simultaneously invites sentencers to flout the Constitution of the United States and promises them that, should they do so, their action is immune from federal judicial review. Astoundingly, the Court in upholding the procedure explicitly commends this very feature. See ante, at 301-302. I do not think that such a procedure is consistent with the Due Process Clause, and I would accordingly reverse petitioner McGautha's sentence of death.

I have indicated above the reasons why, in my judgment, the procedures adopted by Ohio and California to sentence convicted defendants to die are inconsistent with the most basic and fundamental principles of due process. But even if I thought these procedures adequate to try a welfare claim-which they are not, Goldberg v. Kelly, 397 U.S. 254, 255, 90 S.Ct. 1011, 1014, 25 L.Ed.2d 287 (1970)-I would have little hesitation in finding them inadequate where life itself is at stake. For we have long recognized that the degree of procedural regularity required by the Due Process Clause increases with the importance of the interests at stake. See Cafeteria and Restaurant Workers Union Local 473 v. McElroy, 367 U.S. 886, 895-896, 81 S.Ct. 1743, 1748 1749, 6 L.Ed.2d 1230 (1961); id., at 900-901, 81 S.Ct. at 1751 1752 (dissent). Where First Amendment interests have been involved we have held the States to stringent procedural requirements indeed. See, e.g., Stanley v. Georgia, 394 U.S. 557, 89 S.Ct. 1243, 22 L.Ed.2d 542 (1969); Freedman v. Maryland, 380 U.S. 51, 85 S.Ct. 734, 13 L.Ed.2d 649 (1965); A Quantity of Books v. Kansas, 378 U.S. 205, 84 S.Ct. 1723, 12 L.Ed.2d 809 (1964); Marcus v. Search Warrant, 367 U.S. 717, 81 S.Ct. 1708, 6 L.Ed.2d 1127 (1961); Speiser v. Randall, 357 U.S. 513, 78 S.Ct. 1332, 2 L.Ed.2d 1460 (1958). Of course the First Amendment is 'an interest of transcending value,' id., at 525, 78 S.Ct., at 1342, but so is life itself. Yet the Court's opinion turns the law on its head to conclude, apparently, that because a decision to take someone's life is of such tremendous import, those who make such decisions need not be 'inhibit(ed)' by the safeguards otherwise required by due process of law. Ante, at 208. My belief is to the contrary, and I would hold that no State which determines to take a human life is thereby exempted from the constitutional command that it do so only by 'due process of law.'

Finally, a few words should be said about matters peripherally suggested by these cases. First, these cases do not in the slightest way draw into question the power of the States to determine whether or not to impose the death penalty itself, any more than Giaccio v. Pennsylvania, 382 U.S. 399, 86 S.Ct. 518, 15 L.Ed.2d 447 (1966), involved the power of the State of Pennsylvania to impose criminal punishment on persons who should fire a pistol loaded with blanks at another. Second, these cases do not call upon us to determine whether petitioners' trials were 'fairly conducted' in the way referred to by my Brother BLACK. Ante, at 225. What they do call upon us to determine is whether the Due Process Clause requires the States, in his words, 'to make certain that men would be governed by law, not the arbitrary fiat of the man or men in power,' In re Winship, 397 U.S. 358, 384, 90 S.Ct. 1068, 1083, 25 L.Ed.2d 368 (1970) (dissent), and whether if a State, acting through its jury, applies one standard to determine that one convicted criminal should die, 'the Due Process Clause commands that every trial in that jurisdiction must adhere to that standard.' Id., at 386, 90 S.Ct., at 1084. Third, we are not called upon to determine whether 'the death penalty is appropriate punishment' for the petitioners before us. Ante, at 221. That determination is for the States. The Court, however, apparently believes that the procedures before us are to be upheld because the results in the present cases comport with its own, unarticulated notions of capital sentencing policy. See ibid. This fundamental misapprehension of the judicial function pervades the Court's opinion, which after a single brief mention of the Due Process Clause entirely eschews discussion of the Constitution, and instead speaks only of the considerations upon which it believes the States should rest their capital sentencing policy. Ante, at 196-208.

Finally, I should add that for several reasons the present cases do not draw into question the power of the States that should so desire to commit their criminal sentencing powers to a jury. For one thing, I see no reason to believe that juries are not capable of explaining, in simple but possibly perceptive terms, what facts they have found and what reasons they have considered sufficient to take a human life. Second, I have already indicated why I believe that life itself is an interest of such transcendent importance that a decision to take a life may require procedural regularity far beyond a decision simply to set a sentence at one or another term of years. Third, where jury sentencing involves such a decision, determination of the ultimate question-how many years a defendant will actually serve-is generally placed very substantially in the hands of a parole board a single, continuing board of professionals whose general supervision and accumulated wisdom can go far toward insuring consistency in sentencing. And finally, in most cases where juries are asked to fix a convicted defendant's sentence at one or another term of years, they must inevitably be aware that, no matter what they do, the defendant will eventually return to society. With this in mind, a jury should at the very least recognize that rehabilitation must be a factor of substantial weight in its deliberations. Of course, none of these cases are before us, and I do not mean to imply that any and every question other than the question of life or death may be submitted by a State to a jury to be determined in its unguided, unreviewed, and unreviewable discretion. But I cannot help concluding that the Court's opinion, at its core, rests upon nothing more solid than its inability to imagine any regime of capital sentencing other than that which presently exists. I cannot assent to such a basis for decision. 'If we would guide by the light of reason, we must let our minds be bold.' New State Ice Co. v. Liebmann, 285 U.S. 262, 311, 52 S.Ct. 371, 387, 76 L.Ed. 747 (1932) (Brandeis, J., dissenting).