We consider this case en banc to address an issue important to the administration of justice in the federal courts. The question is the constitutionality of the section of the Federal Magistrate Act of 1979 which allows magistrates to conduct civil trials with *540 the consent of all parties. 28 U.S.C. § 636 (c) (Supp. V 1981).

Pacemaker, Inc., brought this suit alleging patent infringement against Instromedix, Inc. Instromedix counterclaimed for a declaration of the patent’s invalidity. The parties, pursuant to 28 U.S.C. § 636 (c) and local rules of the District of Oregon, elected to try the case before a United States Magistrate. The magistrate held the patent valid, but not infringed. Both parties appealed, and a panel of this court, sua sponte, raised the question of the constitutionality of trial by a magistrate, an officer not accorded the protections of Article III, section 1, of the Constitution. The panel held the statute invalid and vacated the judgment, Pacemaker Diagnostic Clinic v. Instromedix, Inc., 712 F.2d 1305 (9th Cir.1983), and the court has convened en banc to reconsider the case. Pacemaker Diagnostic Clinic v. Instromedix, Inc., 718 F.2d 971 (9th Cir. 1983) (order granting rehearing en banc). We hold that, in light of the statutory precondition of voluntary litigant consent and the provisions for the appointment and control of the magistrates by Article III courts, the conduct of civil trials by magistrates is constitutional.

Commencing with the Federal Magistrates Act of 1968, Pub.L. No. 90-578, 82 Stat. 1107, Congress enacted a series of statutes to “reform the first echelon of the Federal judiciary into an effective component of a modern scheme of justice.” H.R. Rep. No. 1629, 90th Cong., 2d Sess. 11, reprinted in 1968 U.S.Code Cong. & Ad.News 4252, 4253-54. The 1968 Act replaced the former office of United States Commissioner with the new office of United States Magistrate. The 1968 Act was designed to improve the quality of these officers and to enlarge their responsibilities. The Act established minimum qualifications for the office, and vested the appointment authority with the judges of each district. 28 U.S.C. § 631 (1976). Magistrates’ jurisdiction was expanded to include duties such as “service as special masters, supervision of pretrial or discovery proceedings, and preliminary consideration of petitions for post-conviction relief.” H.R.Rep. No. 1629, 90th Cong., 2d Sess. 11, reprinted in 1968 U.S. Code Cong. & Ad.News at 4254. See 28 U.S.C. § 636 (a), (b) (1976). Magistrates were also given jurisdiction over minor criminal offenses when the accused waives trial by judge before an Article III court. 18 U.S.C. § 3401 (1976).

The Act was amended in 1976, Pub.L. No. 94-577, 90 Stat. 2729, in response to a number of court decisions which had, as a matter of statutory interpretation, limited the jurisdiction of magistrates. E.g., Wingo v. Wedding, 418 U.S. 461 , 94 S.Ct. 2842 , 41 L.Ed.2d 879 (1974); see S.Rep. No. 625, 94th Cong., 2d Sess. 3-4 (1976); H.R.Rep. No. 1609, 94th Cong., 2d Sess. 5-6, reprinted in 1976 U.S.Code Cong. & Ad.News 6162, 6164-66. By clarifying the statutory language, Congress solidified the jurisdiction of magistrates over certain matters. See McCabe, The Federal Magistrate Act of 1979, 16 Harv.J.Legis. 343, 353-54 (1979).

The section at issue here, 28 U.S.C. § 636 (c), was enacted in 1979 together with other provisions. The Federal Magistrate Act of 1979, Pub.L. No. 96-82, 93 Stat. 643, § 2(2). Congress’ object was “to amend the current jurisdictional provisions for U.S. magistrates ... in order to further clarify and expand the jurisdiction of U.S. magistrates and improve access to the Federal courts for the less-advantaged.” S.Rep. No. 74, 96th Cong., 1st Sess. 1, reprinted in 1979 U.S.Code Cong. & Ad.News 1469, 1469. Thus, section 636(c) authorizes magistrates, when specially designated by the district court, to exercise jurisdiction over civil matters and enter a final judgment in the district court in civil cases, provided the parties consent to the reference.

The challenge to the statute is based on the implicit command of Article III that the judicial power of the United States is confined to judges holding commissions under and cloaked with the protections of that article, and the corollary principle of separation of powers. Article III is one of the provisions of the Constitution which delineates the separation of powers among *541 the branches of government. The attributes of Article III judges, permanency in office and the right to an undiminished compensation, are as essential to the independence of the judiciary now as they were when the Constitution was framed. See The Federalist Nos. 78 & 79 (A. Hamilton); Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U.S. 50, 58-60 , 102 S.Ct. 2858, 2865-66 , 73 L.Ed.2d 598 (1982); United States v. Will, 449 U.S. 200, 217-221 , 101 S.Ct. 471, 481-84 , 66 L.Ed.2d 392 (1980). In addition to the unimpeached precedent supporting this proposition, our own experience attests to the substance and reality of the guarantees. A separate and independent judiciary, and the guarantees that assure it, are present constitutional necessities, not relics of antique ideas.

We have observed that separation of powers protections, in some cases, have two components. One axis reaches to the person affected by government action and encompasses his or her relation to a constitutional branch; the other axis runs from each governmental branch to the others to insure separation and independence in the constitutional structure. Chadha v. INS, 634 F.2d 408, 422, 431 (9th Cir.1980), aff’d, — U.S. —, 103 S.Ct. 2764 , 77 L.Ed.2d 317 (1983). These two aspects of the separation of powers rule are applicable here. Where a case is transferred or assigned from an Article III court to a different forum, both the rights of the parties and the relations between the separate branches of the government are implicated. First, we consider whether transfer of the case to a magistrate invades rights personal to the litigants. Second, we examine whether, even if the parties have consented to the procedure, the existence or operation of the alternate forum compromises the essential independence of the judiciary.

At the outset, and leaving aside all consideration of criminal cases, we recognize the principle that parties to a case or controversy in a federal forum are entitled to have the cause determined by Article III judges, with some significant exceptions yet to be fully delineated by the Supreme Court. Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U.S. 50 , 102 S.Ct. 2858 , 73 L.Ed.2d 598 (1982) (plurality opinion); id. at 89 , 102 S.Ct. at 2881 (Rehnquist, J., concurring). We decide the case before us on the further premise that the patent suit here falls within none of the recognized or potential exceptions, even the public rights exception. Id. at 67-76 , 102 S.Ct. at 2869-2874 . Though no authority we have found expressly so holds, it is appropriate to address the constitutional issue in this patent case. The statutory provision for reference to magistrates applies, without qualification, to all civil cases, and the issue will arise in other matters that do not arguably qualify as a case involving federally created rights. In its recent decision upholding the statute, the Third Circuit adopted the same approach to reach the constitutional issue. See Wharton-Thomas v. United States, 721 F.2d 922, 930 (3d Cir.1983).

The independent character of federal adjudication under Article III imparts to a judgment qualities of authority and respect that are well understood. It follows that the federal litigant has a personal right, subject to exceptions in certain classes of cases, to demand Article III adjudication of a civil suit. Authorities support the premise that Article III adjudication is, in part, a personal right of the litigant. Glid-den Co. v. Zdanok, 370 U.S. 530, 536 , 82 S.Ct. 1459, 1465 , 8 L.Ed.2d 671 (1962) (“The alleged defect of authority here relates to basic constitutional protections designed in part for the benefit of litigants”); Palmore v. United States, 411 U.S. 389, 412 , 93 S.Ct. 1670, 1684 , 36 L.Ed.2d 342 (1973) (Douglas, J., dissenting) (“The safeguards accorded Art. Ill judges were designed to protect litigants with unpopular or minority causes or litigants who belong to despised or suspect classes”); Kurland, The Constitution and the Tenure of Federal Judges: Some Notes from History, 36 U.Chi.L.Rev. 665, 698 (1969) (life tenure of federal judges “not created for the benefit of the judges but for the benefit of the judged”); Kaufman, Chilling Judicial Independence, 88 Yale L.J. 681 , 690 (1979) (same). And it *542 could not be otherwise, for in deciding cases or controversies, the judicial branch acts primarily on the litigants before the court. O’Donoghue v. United States, 289 U.S. 516 , 532, 53 S.Ct. 740, 743 , 77 L.Ed. 1356 (1933) (quoting Chief Justice Marshall in the debates on the Virginia State Convention of 1829-1830).

Magistrates are not protected by Article III status. Their term of office is for four or eight years; they can be removed by the judges of the district or, in some cases, the circuit council; and their salary may be diminished by act of Congress. 28 U.S.C. §§ 631 (e), (i), 633(c) & 634 (1976 & Supp. V 1981). A mandatory provision for trial of an unrestricted class of civil cases by a magistrate and not by Article III judges would violate the constitutional rights of the litigants. Nevertheless, as this aspect of the separation of powers doctrine embodied in Article III is personal to the parties, it may be waived.

In recent cases, the Supreme Court has not had to consider the constitutional implications of consent by the parties to go before a non-Article III judge, because in none did all parties consent to such a procedure. See, e.g, Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U.S. 50 , 102 S.Ct. 2858 , 73 L.Ed.2d 598 (1982); United States v. Raddatz, 447 U.S. 667 , 100 S.Ct. 2406 , 65 L.Ed.2d 424 (1980); Palmore v. United States, 411 U.S. 389 , 93 S.Ct. 1670 , 36 L.Ed.2d 342 (1973); Glidden Co. v. Zdanok, 370 U.S. 530 , 82 S.Ct. 1459 , 8 L.Ed.2d 671 (1962). Significantly, in the Court’s latest interpretation of Article III, all the Justices, although not forced to confront the issue squarely, indicated that consent is important to the constitutional analysis. See Northern Pipeline, 458 U.S. at 79, n. 31 , 102 S.Ct. at 2876, n. 31 (Brennan, J.) (plurality opinion) (“[b]efore the [Bankruptcy Reform] Act the referee had no jurisdiction, except with consent, over controversies beyond those involving property in the actual or constructive possession of the court” (emphasis added)); id. at 91 , 102 S.Ct. at 2882 (Rehnquist, J., concurring) (“I would, therefore, hold so much of the Bankruptcy Act of 1978 as enables a Bankruptcy Court to entertain and decide Northern’s lawsuit over Marathon’s objection to be violative of Art. III of the United States Constitution” (emphasis added)); id. at 92 , 102 S.Ct. at 2882 (Burger, C.J., dissenting) (“the Court’s holding is limited to the proposition stated by Justice Rehnquist in his concurrence in the judgment — that a ‘traditional’ state common-law action, not made subject to a federal rule of decision, and related only peripherally to an adjudication of bankruptcy under federal law, must, absent the consent of the litigants, be heard by an ‘Article III court’” (emphasis added)); id. at 95 , 102 S.Ct. at 2884 (White, J., dissenting) (under the majority’s interpretation, state law claims “would have to be heard by Art. Ill judges or by state courts — unless the defendant consents to suit before the bankruptcy judge — just as they were before the 1978 Act was adopted” (emphasis added)). See also McCabe, The Federal Magistrate Act of 1979, 16 Harv.J.Legis. 343, 374-79 (1979); Silberman, Masters and Magistrates Part II: The American Analogue, 50 N.Y.U.L.Rev. 1297, 1350-54 (1975); Comment, An Adjudicative Role for Federal Magistrates in Civil Cases, 40 U.Chi.L.Rev. 584 (1973) (all arguing that litigant consent solves any Article III constitutional problems under the new Magistrate Act). The Third Circuit has concluded that consent under § 636(c) cures any constitutional defects, Wharton-Thomas v. United States, 721 F.2d 922, 925-926 (3d Cir.1983), and that conclusion is entitled to deference.

We also give considerable weight to the judgment of Congress that consent of the parties eliminates constitutional objections. The House Committee gave explicit consideration to the issue of constitutionality, and concluded that consent of the parties suffices to overcome objections based on constitutional grounds. Compare H.R. Rep. No. 287, 96th Cong., 1st Sess. 7-9 (1979) (House Committee confident that the Magistrate Act passes constitutional muster) with H.R.Rep. No. 595, 95th Cong. 1st Sess., 36-39 (1977), reprinted in 1978 U.S. *543 Code Cong. & Ad.News, 5787, 5963, 5997-6000; and, Staff of House Subcommittee on Civil and Constitutional Rights, Committee on the Judiciary, 95th Cong., 1st Sess., Constitutional Bankruptcy Courts 33 (Committee Print 1977) (same Committee expressing substantial doubts about the constitutionality of the Bankruptcy Reform Act the Supreme Court subsequently struck down in Northern Pipeline).

The Supreme Court has allowed criminal defendants to waive even fundamental rights: the right to be free from self-incrimination, Garner v. United States, 424 U.S. 648 , 96 S.Ct. 1178 , 47 L.Ed.2d 370 (1976), the right to counsel, Adams v. United States ex rel. McCann, 317 U.S. 269 , 63 S.Ct. 236 , 87 L.Ed. 268 (1942), the right to be free from unreasonable searches and seizures, Schneckloth v. Bustamonte, 412 U.S. 218 , 93 S.Ct. 2041 , 36 L.Ed.2d 854 (1973), the right to a speedy trial, Barker v. Wingo, 407 U.S. 514 , 92 S.Ct. 2182 , 33 L.Ed.2d 101 (1972), the right to a jury trial, Duncan v. Louisiana, 391 U.S. 145, 158 , 88 S.Ct. 1444, 1452 , 20 L.Ed.2d 491 (1968), and even, by pleading guilty, the right to trial itself. See Boykin v. Alabama, 395 U.S. 238, 243 , 89 S.Ct. 1709, 1712 , 23 L.Ed.2d 274 (1969). We refuse to reach the anomalous result of forbidding waiver in a civil case of the personal right to an Article III judge.

The waiver of personal rights must, of course, be freely and voluntarily undertaken. The present statute recognizes the importance of this requirement and provides that “[r]ules of court for the reference of civil matters to magistrates shall include procedures to protect the voluntariness of the parties’ consent.” 28 U.S.C. § 636 (c)(2). The local rules of the District of Oregon pertaining to the assignment of civil cases to magistrates meet this requirement.

The purported waiver of the right to an Article III trial would not be an acceptable ground for avoiding the constitutional question if the alternative to the waiver were the imposition of serious burdens and costs on the litigant. If it were shown that the choice is between trial to a magistrate or the endurance of delay or other measurable hardships not clearly justified by the needs of judicial administration, we would be required to consider whether the right to an Article III forum had been voluntarily relinquished. No such burdens or hardships have been demonstrated here. Access to district judges in the current judicial system is not so restricted that adjudication of cases by magistrates is a compelled alternative.

Pacemaker argues that in the federal system a party may not consent to jurisdiction, so that the parties cannot waive their rights under Article III. The maxim that parties may not consent to the jurisdiction of federal courts is not applicable here. The rule is irrelevant because it applies only where the parties attempt to confer upon an Article III court a subject matter jurisdiction that Congress or the Constitution forbid. See, e.g, Jackson v. Ashton, 33 U.S. (8 Peters), 148, 148-49, 8 L.Ed. 898 (1834); Mansfield, Coldwater & Lake Michigan Railway Co. v. Swan, 111 U.S. 379 , 4 S.Ct. 510 , 28 L.Ed. 462 (1884). The limited jurisdiction of the federal courts and the need to respect the boundaries of federalism underlie the rule. In the instant ease, however, the subject matter, patents, is exclusively one of federal law. The Supreme Court has explicitly held that Congress may “confer upon federal courts jurisdiction conditioned upon a defendant’s consent.” Williams v. Austrian, 331 U.S. 642, 652 , 67 S.Ct. 1443, 1448 , 91 L.Ed. 1718 (1947); see Harris v. Avery Brundage Co., 305 U.S. 160 , 59 S.Ct. 131 , 83 L.Ed. 100 (1938). The litigant waiver in this case is similar to waiver of a defect in jurisdiction over the person, a waiver federal courts permit. Hoffman v. Blaski, 363 U.S. 335, 343 , 80 S.Ct. 1084, 1089 , 4 L.Ed.2d 1254 (1960).

The maxim also begs the essential question, whether jurisdiction can be conferred on the magistrate. The issue is not the expansion of Article III jurisdiction but its transfer to another federal forum. The component of the separation of powers rule that protects the integrity of the constitutional structure, as distinct from the compo *544 nent that protects the rights of the litigants, cannot be waived by the parties, but that analysis requires us to address questions more fundamental than the irrelevant rule that jurisdiction cannot be conferred by the parties. We now address those concerns.

On its most fundamental plane, the separation of powers doctrine protects the whole constitutional structure by requiring that each branch retain its essential powers and independence. Buckley v. Valeo, 424 U.S. 1, 120 , 96 S.Ct. 612, 682-83 , 46 L.Ed.2d 659 (1976); Chadha v. INS, 634 F.2d at 425 (9th Cir.1980), aff’d, — U.S. —, 103 S.Ct. 2764 , 77 L.Ed.2d 317 (1983). Statutes or governmental actions which violate the separation of powers doctrine in its systemic aspect should be invalidated, as a general rule, despite waiver by affected private parties. Here the statute requires entry and enforcement of judgments in an Article III court after a non-Article III judicial officer has heard the suit. The procedure raises at least two concerns for the integrity of the judiciary. The first is whether, by providing for reference of court cases to a magistrate, Congress has invaded the power of a coordinate branch or permitted an improper abdication of that branch’s central authority. See, e.g., Buckley v. Valeo, 424 U.S. 1, 120-43 , 96 S.Ct. 612, 682-94 , 46 L.Ed.2d 659 (1976); Schechter Poultry Corp. v. United States, 295 U.S. 495, 529-42 , 55 S.Ct. 837, 842-48 , 79 L.Ed. 1570 (1935); the second is whether the requirement for entry of judgment improperly directs the judiciary in the performance of its duties. See United States v. Klein, 80 U.S. (13 Wall.) 128 , 20 L.Ed. 519 (1871); cf. California Medical Ass’n v. Federal Election Comm'n, 641 F.2d 619, 631-32 (9th Cir.1980) (en banc) (statutory provision requiring en banc consideration of cases arising under Federal Election Campaign Act may present constitutional questions), aff'd, 453 U.S. 182 , 101 S.Ct. 2712 , 69 L.Ed.2d 567 (1981).

The standard for determining whether there is an improper interference with or delegation of the independent power of a branch is whether the alteration prevents or substantially impairs performance by the branch of its essential role in the constitutional system. Nixon v. Administrator of General Services, 433 U.S. 425, 443 , 97 S.Ct. 2777, 2790 , 53 L.Ed.2d 867 (1977). There is little guidance in the cases beyond this broad formulation, but the general rule can be narrowed for the case before us. If the essential, constitutional role of the judiciary is to be maintained, there must be both the appearance and the reality of control by Article III judges over the interpretation, declaration, and application of federal law. See Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U.S. 50, 76-81 , 102 S.Ct. 2858, 2874-77 , 73 L.Ed.2d 598 (1982); United States v. Raddatz, 447 U.S. 667, 685 , 100 S.Ct. 2406, 2417 , 65 L.Ed.2d 424 (1980) (Blackmun, J., concurring). The required control must be more than simple appellate review. This is the teaching of Northern Pipeline. 458 U.S. at 86, n. 39 , 102 S.Ct. at 2879, n. 39 .

Ours is not the paradigmatic separation of powers case, where the integrity of one branch is threatened by another which attempts an arrogation of power to itself. E.g, Buckley v. Valeo, 424 U.S. 1, 120 , 96 S.Ct. 612, 682-83 , 46 L.Ed.2d 659 (1976); Youngstown Steel & Tube v. Sawyer, 343 U.S. 579 , 72 S.Ct. 863 , 96 L.Ed. 1153 (1952). The potential for disruption is instead the erosion of the central powers of the judiciary by permitting it to delegate its own authority. Upon examination of the statute before us, we conclude that it contains sufficient protection against the erosion of judicial power to overcome the constitutional objections leveled against it. The statute invests the Article III judiciary with extensive administrative control over the management, composition, and operation of the magistrate system. It permits, moreover, control over specific cases by the resumption of district court jurisdiction on the court’s own initiative.

Article III courts control the magistrate system as a whole. The statutory scheme created by. Congress protects against intervention by political branches of govern *545 ment. The Judicial Conference of the United States, composed exclusively of Article III judges, see 28 U.S.C. § 331 (1976), determines the number of magistrate positions for each district, 28 U.S.C. § 633 (b), protecting against the designation of so many magistrates that effective judicial control is lost. We must acknowledge that the exclusive character of Article III control in this one aspect of the statute will lose considerable force under an amendment effective April 1984, when bankruptcy judges, non-Article III officers, are included within the Judicial Conference of the United States. 28 U.S.C. § 331 (Supp. V 1981). There are enough other aspects of exclusive judicial control so that this change should not alter our conclusion, though it does detract from the constitutional symmetry of the Magistrate Act.

Selection of magistrates and their retention in office, under statutory standards relating to performance and fitness, is the responsibility of Article III judges, as it is vested in the judges of the separate districts. 28 U.S.C. § 631 . Magistrates thus are not made directly dependent upon loyalty to officers in either of the political branches.

The congressional designation of Article III judges to select and to appoint magistrates as subordinate officers is not the mark of an aberrant procedure. To the contrary, the Constitution contains explicit, textual authority for the judiciary to appoint its own officers, if the Congress so permits. Article II, section 2 provides that “the Congress may by law vest the appointment of such inferior officers, as they think proper, in the President alone, in the Courts of Law, or in the Heads of Departments.” See Go-Bart Importing Co. v. United States, 282 U.S. 344, 352-53 , 51 S.Ct. 153, 156-57 , 75 L.Ed. 374 (1931) (commissioners are inferior officers); Rice v. Ames, 180 U.S. 371, 378 , 21 S.Ct. 406, 409 , 45 L.Ed. 577 (1901) (Congress may authorize judges to appoint commissioners). This constitutional authority for the exercise of the appointment power by Article III judges implies an important dimension to the judicial power: the judiciary is permitted a degree of control and discretion for the design and shape of its own system. The Magistrates Act implements this constitutional authority.

In addition to judicial supervision over the magistrate system as a whole, there is a further element of judicial control in the Article III authority to cancel an order of reference, sua sponte or on application of the parties, in individual cases. 28 U.S.C. § 636 (c)(6). The district court’s power to void a reference sua sponte is predicated upon good cause, a term yet to be explored in the context of specific cases. It would seem at a minimum, however, that good cause for resumption of direct Article III control exists in a case where a political branch of the government is directly affected, or where a substantial constitutional question is presented, or where rights of numerous parties not present before the court might be affected by the decision, or in any other case containing sensitivities such that determination by an Article III judge is required to insure the appearance and the reality of independence and impartiality in the decision. See S.Rep. No. 74, 96th Cong., 1st Sess. 14, reprinted in 1979 U.S.Code Cong. & Ad.News 1469, 1483. By contrast, certain early Supreme Court cases, while not explicitly addressed to separation of powers concerns, support the proposition that consensual reference to non-Article III judges is permitted in some types of cases. E.g., Kimberly v. Arms, 129 U.S. 512 , 9 S.Ct. 355 , 32 L.Ed. 764 (1889) (ownership interest in stock); Newcomb v. Wood, 97 U.S. (7 Otto) 581, 24 L.Ed. 1085 (1878) (restitution for goods sold to bankrupt); Heckers v. Fowler, 69 U.S. (2 Wall.) 123 , 17 L.Ed. 759 (1864) (breach of a patent license and defense of invalidity of patent); Alexandria Canal Co. v. Swann, 46 U.S. (5 How.) 83 , 12 L.Ed. 60 (1847) (trespass to land).

Article III authority is preserved in other respects. District courts retain the power to adjudge a party in contempt. 28 U.S.C. § 636 (e). Sections 636(c)(3) and (4) provide for appellate review of the judgment of the magistrate by the appropriate court of appeals as a matter of right or, if *546 the parties consent, by the district court. 28 U.S.C. §§ 636 (c)(3), (4). The Act imposes no limits on review by the Supreme Court. 28 U.S.C. §§ 636 (c)(3), (5). Article III courts retain full authority over questions of law.

The power to cancel a reference, taken together with the retention by Article III judges of the power to designate magistrate positions and to select and remove individual magistrates, provides Article III courts with continuing, plenary responsibility for the administration of the judicial business of the United States. This responsibility sufficiently protects the judiciary from the encroachment of other branches to satisfy the separation of powers embodied in Article III.

It is instructive to compare and contrast reference of civil cases to a magistrate under § 636(c), the provision at issue here, with the statutory scheme providing for de novo review of a magistrate’s determination under § 636(b), the system upheld by the Supreme Court in United States v. Rad-datz, 447 U.S. 667 , 100 S.Ct. 2406 , 65 L.Ed.2d 424 (1980). We must recognize that neither section insures full Article III participation in the decision of every case. Each has its own potential defects and comparative advantages. The merit of the de novo determination scheme lies in the duty of the district court to make its own determinations of magistrate rulings to which the parties object, 28 U.S.C. § 636 (b)(1); but the defect inherent is the lack of assurance that the full power of the district court will be exercised to rehear the evidence on credibility issues that may be central in the case. See United States v. Rad-datz, 447 U.S. at 683 & n. 11, 100 S.Ct. at 2416 & n. 11. Under the de novo determination rule, the district court may give some deference to the findings of magistrates. Id. The plan contains at least the potential for district judges to engage in routine and cursory examinations, so that decisions nominally made by district judges will in fact be those of the magistrate. The identity of the officer responsible for the decision may be unclear to the public or to the litigants. Where the responsibility becomes blurred, power may be either abused or too timidly exercised. See Note, Article III Constraints and the Expanding Civil Jurisdiction of Federal Magistrates: A Dissenting View, 88 Yale L.J. 1023 , 1047-58 (1979).

The consensual case reference scheme, before us here, provides a much better delineation of the decisional process. There should be little danger that litigants or the public will be confused over whether the magistrate or the district judge has decided the case. The principal disadvantage of the consensual reference plan may arise in the tendency to overuse it. Even if the number of magistrates and case referrals become substantially greater than it is now, there must be active supervision by Article III judges. The statutory safeguards are the power of the district judges to cancel references for good cause, and the requirement of litigant consent. 28 U.S.C. §§ 636 (c)(1), (2), (6). Continued and vigilant supervision by Article III judges is of course essential to the integrity of the system, and they must be careful to guard against any compulsion to induce consent through the imposition of costs, delays, or other penalties which would be incompatible with the foregoing conclusion that the consent of the parties is essential to the constitutionality of the Act.

At this stage in the evolution of the magistrate system, however, we cannot say the Congress was constitutionally compelled to rely on the de novo determination plan approved in Raddatz to the exclusion of the consensual case reference system under examination here. There are adequate protections in the statute so that it is not invalid on its face or as applied in the present structure of the judicial system.

From a realistic and practical perspective, reference of civil cases to magistrates with the consent of the parties, subject to careful supervision by Article III judges, may serve to strengthen an independent judiciary, not undermine it. The Constitution is not simply an exercise in metaphysics. The idea of separation of *547 powers is justified by eminently practical considerations. See The Federalist No. 51 (J. Madison) at 349 (J. Cooke ed. 1961); Chadha v. INS, 634 F.2d at 423-24 . It is faithful to the idea of separation of powers to examine the real consequences of the statute.

There are compelling reasons for the creation of an infrastructure for determining certain civil cases with the consent of the parties and subject to judicial control. Article III courts have the task of adjudicating an ever-mounting volume of cases, see Administrative Office of the United States Courts, Federal Court Management Statistics 1983 (1983). Further, the entire character of litigation that federal courts principally face is changing. Chayes, The Supreme Court 1981 Term— Foreword: Public Law Litigation and the Burger Court, 96 Harv.L.Rev. 4 (1982). The legislature and the judiciary act responsibly when they provide and explore new, flexible methods of adjudication, especially where the evolution of the innovative mechanism is left in large part under the control of the judiciary itself. The Magistrate Act was shaped by these considerations. See S.Rep. No. 74, 96th Cong., 1st Sess. 4-5, reprinted in 1979 U.S.Code Cong. & Ad. News 1469, 1472-74.

The patent dispute here required an explanation of the elements of design and the mechanics of a complex computer system used to monitor the performance of artificial implants for sustaining the heartbeat. While we have not examined thé validity of the legal conclusions drawn by the magistrate, in the course of a 28 page opinion he gave a meticulous explanation of these matters. The procedure is hardly an affront to the constitutional system where the parties have consented to it. In these circumstances, the case is not one necessarily calling for adjudication by a judge with the protections of Article III.

We hold that consensual reference of a civil case to a magistrate is constitutional, and we remand to the three judge panel of this court to decide the substantive issues of patent law on appeal.

SCHROEDER, Circuit Judge, with whom PREGERSON and CANBY, Circuit Judges, join, dissenting.

U.S. Const, art. Ill, § 1.

The majority holds that, so long as the parties consent, the power to decide any federal civil case in the district courts of the United States may be exercised by individuals who occupy positions that Congress has not established, and who must depend both upon district judges for their tenure and upon Congress for their compensation. Because I believe this holding disrupts the proper operation of our constitutional system, including the independent exercise of judicial power by individuals free of outside constraints, I dissent.

The majority bases its holding on three well-meaning but fundamentally misguided assumptions. The first assumption is that under our constitutional system the exercise of the judicial power of the United States by Article III judges can depend upon stipulations of the litigants. The second is that magistrates, who operate under the thumb of district court judges and whose salaries are not protected from retaliatory diminution by Congress, have the independence the Constitution is designed to ensure. The third is that consent to use of a magistrate can be presumed to be voluntary when the explicit purpose of the consensual reference provision of the Magistrates Act was to encourage certain classes of litigants to abandon their right to Article III adjudication because existing overburdened district judges could not hear all cases promptly-

The fallacies inherent in these assumptions become apparent after examination of *548 the effect the reference provision has upon the relationship among the three branches of government, the effect it has upon the independence of those who decide cases, and its effect upon litigants’ right to adjudication by Article III judges.

I. The Effect of Consensual Reference to Magistrates Upon Our Constitutional System

The Magistrates Act, 28 U.S.C. §§ 631-639 (1976 and Supp. V 1981), authorizes the judicial branch to create new judicial offices, appoint those who will occupy the offices, and then delegate to appointees the authority to exercise the civil judicial power of the United States. 28 U.S.C. § 636 (c). The Act creates mutations in our system of government that transcend its effects on individual litigants.

Ours is a system of three separate departments of government, each exercising cheeks on the others. In the words of James Madison:

The Federalist No. 48, at 332 (J. Cooke ed. 1961).

The judiciary is the principal check on the usurpation of power by the other branches. Judges are to be what Madison described as the “expositors of the Laws.” 2 Records of the Federal Convention of 1787 IS (M. Farrand rev. ed. 1937). As Chief Justice Marshall pronounced, “It is, emphatically, the province and duty of the judicial department, to say what the law is.” Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 , 2 L.Ed. 60 (1803). Early legislators recognized that “[s]o long as we may have an independent judiciary, the great interests of the people will be safe.” 11 Annals of Cong. 739-40 (1802) (statement of John Rutledge, Jr.).

The judiciary’s unique role in the tripartite system is to interpret both the Constitution, which is the fundamental law, and the acts of Congress:

The Federalist No. 78, at 525 (A. Hamilton) (J. Cooke ed. 1961). The independence of the judiciary was essential to Hamilton. In the same paper he wrote:

Id. at 524.

Justice Story’s commentaries emphasized the importance of independent judges for the protection of minorities from the tyranny of the majority.

2 J. Story, Commentaries on the Constitution of the United States 403 (1873).

Article Ill’s provisions for tenure in office and undiminishable salary were meant *549 to ensure the independence of Article III judges. Hamilton described tenure as the “best expedient ... to secure a steady, upright and impartial administration of the laws.” The Federalist No. 78, at 522 (J. Cooke ed. 1961). The salary provisions, in Hamilton’s view, were next in importance for the independence of the judiciary. The Federalist No. 79, at 531 (J. Cooke ed. 1961). Also important was the provision limiting removal of judges to impeachment for misbehavior by the House and trial by the Senate. “This is the only provision on the point, which is consistent with the necessary independence of the judicial character ....” Id. at 533.

The Framers thus built into the Constitution the formula for judicial independence because the role of the judiciary was to be so central to the maintenance of our system of government. They “sought to make the federal judges servants ... only of their consciences.” United States v. Woodley, 726 F.2d 1328, at 1332 (9th Cir.1983). Under these standards, magistrates are not independent. They are beholden to the Article III judiciary for their appointment, retention, and authority to decide cases. They are beholden to Congress for their pay.

Yet the Magistrates Act contemplates that magistrates will exercise the judicial power, acting as the “expositors of the laws” applicable to us all. Under the Act magistrates may, among their many judicial duties, review the constitutionality of actions taken by the other branches. Any failure in execution of this duty injures the entire constitutional system by diminishing the judiciary’s check on the other branches. The exercise of the judicial power by magistrates whose independence is so seriously compromised places our constitutional system at risk.

The loss of the independent exercise of judicial power, the principal check on encroachment by the legislative and executive branches, is not the only effect of the Magistrates Act on our system of government. The Act also interferes seriously with the legislative and executive checks on incursions by the judiciary.

Article III, section 1 of the Constitution vests the judicial power of the United States in one Supreme Court “and in such inferior Courts as the Congress may from time to time ordain and establish.” Article I, section 8, clause 9 directly grants Congress the power “To constitute Tribunals inferior to the Supreme Court.” In the Magistrates Act Congress has delegated to judicial councils the power to create magistrate positions and thereby has abdicated this constitutional responsibility. In practical terms, this abdication reduces the pressure on Congress to create more Article III judgeships, and increases the pressure on district courts to escalate the use of magistrates. Although the Third Circuit, in Wharton-Thomas v. United States, 721 F.2d 922, 930 (3d Cir.1983), recognized this problem forthrightly, it has ignored the constitutional significance of the force creating these practical effects. The majority in this case has ignored even the practical dilemma.

The Magistrates Act also undermines the appointment power of the President and the confirmation power of the Senate under Article II, Section 2. Appointment of magistrates by the judicial council deprives both the President and the Senate of any voice in the selection of individuals who are to exercise Article III powers. This too is a fundamental interference with our system of checks and balances. See the discussion in Buckley v. Valeo, 424 U.S. 1, 120-33 , 96 S.Ct. 612, 682-88 , 46 L.Ed.2d 659 (1976) (Congress impermissibly encroached on President’s appointment power when it gave president pro tempore of Senate power to appoint majority of Federal Election Commission’s voting members).

The majority suggests that Article II, section 2, contemplates some authority in Congress to permit the courts to appoint magistrates as “inferior officers.” The inferior officers clause, however, traditionally has referred to such offices as clerks of court, Ex parte Hennen, 38 U.S. (13 Pet.) 230 , 10 L.Ed. 13 (1839). See Go-Bart Co. v. *550 United States, 282 U.S. 344, 352-53 , 51 S.Ct. 153, 156-57 , 75 L.Ed. 375 (1931) (commissioners may be appointed by district courts as inferior officers because their actions are “preparatory and preliminary” to disposition of case by judge); accord Rice v. Ames, 180 U.S. 371 , 21 S.Ct. 406 , 45 L.Ed. 577 (1901). Such a position is very different from that of the magistrate, who exercises Article III decision making power. Justice Story observed 150 years ago that the issue regarding the appointment of judges had been resolved as a matter of practical construction. Judges of the inferior courts, he said, “are not such inferior officers.” Instead, they are “Officers” who must be appointed by the President with the consent of the Senate. 2 J. Story, supra, at 402 n. 2. No case until this provision was passed has ever challenged this conclusion. 1

The lesson of the Framers is that those who exercise the judicial power of the United States under Article III must be Article III judges. The Supreme Court decisions of the last half-century confirm that consensual reference under the Magistrates Act unconstitutionally authorizes the exercise of Article III power by non-Article III officers because it authorizes the magistrates in civil cases to make final decisions as to all matters of fact and law in any case within the civil jurisdiction of the district courts. 2 Contrary to the assertions of the majority here and in Wharton-Thomas , the parties’ consent does not solve the constitutional problems arising from this wholesale delegation of judicial power to non-Article III judges. Indeed, consent is simply irrelevant to the Supreme Court’s analyses of the proper allocation of judicial power under the Constitution; the judicial power of the United States is conferred upon Article III judges by the Constitution, not by the parties.

In Crowell v. Benson, 285 U.S. 22 , 52 S.Ct. 285 , 76 L.Ed. 598 (1932), the majority upheld use of Unemployment Compensation Commissioners to make factual determinations about claims for injuries on navigable waters. The Court held that “the reservation of full authority to the court to deal with matters of law provides for the appropriate exercise of the judicial function in this class of cases.” Id. at 54 , 52 S.Ct. at 293 . More recently, in upholding on statutory grounds the power of the district courts to refer all Social Security cases to a magistrate under section 636(b) of the Magistrates Act, the majority in Mathews v. Weber, 423 U.S. 261 , 96 S.Ct. 549 , 46 L.Ed.2d 483 (1976), presupposed that the judge makes the final decision. The Court noted that reference of fact-finding duties would improve the judge’s decision. “In this narrow range of cases, reference promotes more focused, and so more careful, decisionmaking by the district judge.” Id. at 274 , 96 S.Ct. at 556.

United States v. Raddatz, 447 U.S. 667 , 100 S.Ct. 2406 , 65 L.Ed.2d 424 (1980), is even more to the point. That case concerned whether the district court was required to hear witnesses in reviewing a magistrate’s suppression motion recommendations under section 636(b) of the Magis *551 trates Act. Chief Justice Burger’s reasoning in the majority opinion, which held that the judge who makes the final decision need not hold a hearing, is telling in the present case:

Id. at 683 , 100 S.Ct. at 2416 (emphasis added), quoting Mathews v. Weber, 423 U.S. at 275 , 96 S.Ct. at 556 ).

Justice Marshall’s dissent in Raddatz viewed the assessment of witness credibility as an essential part of the judicial decision. 447 U.S. at 694 , 100 S.Ct. at 2421 . He echoed the concern of Justice Story, quoted above, that for the protection of minorities those who render judicial decisions must be independent.

Id. at 704 , 100 S.Ct. at 2427 .

In the present case, under either the majority or dissenting analyses in Baddatz, magistrates deciding entire cases exercise the judicial power in violation of constitutional commands.

Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U.S. 50 , 102 S.Ct. 2858 , 73 L.Ed.2d 598 (1982), the most recent Supreme Court decision on the exercise of judicial power by non-Article III judges, compels the same conclusion as Bad-datz. The question in Marathon was whether Congress constitutionally could give non-Article III officers authority to adjudicate a bankruptcy case in which a debtor in Chapter 11 proceedings sued a creditor on breaches of contract and warranty claims. Id. at 56-57 , 102 S.Ct. at 2863-2864 .

Writing for the plurality, Justice Brennan held that Congress’s delegation of power to bankruptcy judges was unconstitutional. He reasoned that Congress could create legislative courts without Article III protections in only three limited settings: territorial courts, courts martial, and courts deciding disputes involving public rights that Congress created in the first instance. Id. at 64-67 , 102 S.Ct. at 2868-2869 . Since the bankruptcy court decided private contests between debtors and creditors, it fit into none of these exceptions and, therefore, it had to be constituted as an Article III court. Id. at 76 , 102 S.Ct. at 2874 . Like bankruptcy courts, magistrates do not fall within Justice Brennan’s three exceptions; magistrate powers extend nationwide to every district court and to all civil eases. For the many reasons discussed thoroughly in the panel opinion in this case, Pacemaker Diagnostic Clinic v. Instromedix, Inc., 712 F.2d 1305 (9th Cir.1983), the plurality reasoning requires a holding that the consensual reference provision in the Magistrates Act is unconstitutional.

Justice White’s dissent in Marathon looked to the scope and significance of the subject matter of the litigation to decide whether Article III judges were required. Applying a balancing test to the bankruptcy court, he focused on bankruptcy as a “specialized [area] having particularized needs and warranting distinctive treatment.” 458 U.S. at 115 , 102 S.Ct. at 2844 (quoting Palmore v. United States, 411 U.S. 389, 407-08 , 93 S.Ct. 1670, 1681-82 , 36 L.Ed.2d 342 (1973)). Further, he noted that the subject matter delegated to the bankruptcy court would be of “little interest to the political branches.” 458 U.S. at 115 , 102 S.Ct. at 2894 .

Consensual reference under the Magistrates Act does not survive Justice White’s balance any more than it survives Justice Brennan’s analysis for the plurality. The reference provision applies to all civil cases, violating Justice White’s concern that Article III give way only in a “specialized area.” *552 The types of cases that can be decided under the reference provision are most certainly of interest to the political branches.

Marathon thus extends the line of cases beginning with Crowell and continuing through Raddatz which, while containing many differences, are all connected by one critical theme: Congress may not delegate away power that properly belongs in the hands of Article III judges. When the power taken from the Article III courts is the power to make final decisions in any type of civil case, as in the Magistrates Act, the reasoning of all of these decisions leads me inescapably to the conclusion that the Constitution has been violated. Just as the Constitution would not permit Congress, with the President’s consent, to usurp the executive power, see Chadha v. INS, 634 F.2d 408, 424 (9th Cir.1980), aff’d, — U.S. —, 103 S.Ct. 2764 , 77 L.Ed.2d 317 (1983), so this court should not sanction any consensual abrogation of power, by Congress and litigants, that the Constitution reserves to the independent judiciary.

II. The Magistrate’s Dependence on the District Court

The majority in this case and the Third Circuit panel in Wharton-Thomas acknowledge that consent of the parties alone does not cure all of the constitutional ills created by transfer of Article III powers to magistrates who lack Article III protections. See majority op. at 543-44; Wharton-Thomas, 721 F.2d at 930 . They conclude, however, that the district judges’ control over magistrates cures any remaining ones. In doing so they focus only upon the threat of erosion of the power exercised by Article III judges and ignore other separation of powers issues. They thus satisfy themselves that the district judges retain their judicial power because district judges maintain, in the words of the majority in this case, “the appearance and the reality of control” over civil cases. Majority op. at 544. That control, we are told, is sufficient, even without statutory guidance as to when and how it is to be exercised, to ensure that withdrawals of references will be made when appropriate. This, according to the majority, is a better system than the district court review of magistrate recommendations approved in Raddatz . In my opinion, this “control” makes a bad matter worse.

The majority assumes that district judges will be able to tell when magistrates are making mistakes. The purpose of the consensual reference provision, however, was to relieve district court judges from some of their decisional responsibilities. If it were possible for district judges to supervise all civil cases to the extent the majority contemplates, there would be no need for magistrates. District judges should be making these decisions in the first instance.

What control by the district court does do is create a severe conflict of interest for the magistrate who is compelled to choose between what the magistrate concludes is right and the result the magistrate thinks will please the district court. Indeed, it is difficult to conceive of greater restraints on the exercise of independent judgment than the provisions in the Act that permit a magistrate to be dismissed, or assignment of a case vacated at any time, by the district court under undefined procedures and malleable standards. 28 U.S.C. § 636 (c)(6) (1976 and Supp. V 1981). If there is a greater disincentive to independent action, it is fear that a particular decision, though legally correct, might lead Congress to lower the pay of all magistrates. That situation, of course, is also possible under this Act.

The constitutional implications of judges controlling other judges outside the appellate process are not unique to this Act. They also have surfaced during the debate of the last twenty years over judicial discipline, removal, and other regulation of conduct by judges.

Strong voices have argued that under our Constitution, no judge should be accountable to any other judge. Judge Irving Kaufman of the Second Circuit, arguing that permitting judges to remove other judges from office frustrates judicial individualism, stated: “Judicial independence, like free expression, is most crucial and most vulnerable in periods of intolerance, when *553 the only hope of protection lies in clear rules setting forth the bright lines that cannot be traversed.” Kaufman, Chilling Judicial Independence, 88 Yale L.J. 681 , 715 (1979). Judge Wallace, of this Circuit, referring to a threat of overzealous programs for judicial efficiency, has written that “it is incumbent upon the judges themselves to fight back” if these programs “interfere with ‘total and absolute independence of judges in deciding cases or in any phase of the decisional function.’ ” Wallace, Judicial Administration in a System of Independents: A Tribe With Only Chiefs, 1978 B.Y.U.L.Rev. 39, 56 (quoting J. Covington, Autonomy v. Efficiency — The Continuing Debate on Judicial Supervision of Federal Trial Judges 43 (July 23, 1973) (unpublished paper)). Professor Kurland declared himself in “wholehearted agreement” with Justice Black’s dissent from the denial of a stay in Chandler v. Judicial Council, 382 U.S. 1003 , 86 S.Ct. 610 , 15 L.Ed.2d 494 (1966) (mem.), which stated:

Chandler, 382 U.S. at 1005-06 , 86 S.Ct. at 611 , quoted in P. Kurland, The Constitution and the Tenure of Federal Judges: Some Notes from History, 36 U.Chi.L.Rev. 665, 667 (1969).

Some Supreme Court consideration of the issue came in Chandler v. Judicial Council, 398 U.S. 74 , 90 S.Ct. 1648 , 26 L.Ed.2d 100 (1970). The recognition of the need to preserve the decision making function independent of the influence of any other judge is reflected in all three opinions in the case. The majority denied a district judge’s petition for extraordinary relief from a judicial council order, in which the district judge had acquiesced, withdrawing case assignments. The Chief Justice wrote: “There can, of course, be no disagreement among us as to the imperative need for total and absolute independence of judges in deciding the cases or in any phase of the decisional function.” Id. at 84 , 90 S.Ct. at 1653 . Justice Harlan’s concurrence, as well, noted the basic responsibility of judges for “final adjudication of law suits.” Id. at 110 , 90 S.Ct. at 1667 . Justice Douglas’ dissent sounded a stronger chord: “Once a federal judge is confirmed by the Senate and takes his oath, he is independent of every other judge. He commonly works with other federal judges who are likewise sovereign.” Id. at 136 , 90 S.Ct. at 1680 .

The majority in the present case, without reference to any of the concerns expressed by all of these judges and scholars, has concluded that the provisions in the Magistrates Act for district judges to control the appointment of magistrates, their reappointment, their assignment of cases, and even the withdrawal of assignments, somehow preserve the independence of decision making. I submit that in reality, the control by the district judges prevents it. If judges are to possess the freedom from political influences of any kind necessary to ensure their ability to decide cases impartially, then, in addition to the express protections of Article III, review of their actions must occur only in accordance with appeal procedures developed under standards of due process.

III. The Illusion of Voluntary Consent

The majority acknowledges that consent to trial by magistrate must be voluntary to avoid violating the litigants’ rights to an Article III judge. Without citation of authority and drawing only upon the absence of any complaint of involuntariness in the present ease, the majority assumes that consent can always be the product of free choice. This assumption ignores the practical realities behind the Magistrates Act’s passage and the very real pressures on dis *554 trict judges to try to channel more and more cases to magistrates.

Congress perceived the Act as a means to cope with increasingly crowded federal court dockets. The legislative history demonstrates that Congress recognized that greater availability of magistrates would induce economically disadvantaged litigants, unable to afford the delay and cost of waiting for adjudication by an Article III judge, to consent to trial before a magistrate. In a Senate Report on the 1979 Act, the Committee on the Judiciary stated:

S.Rep. No. 74, 96th Cong., 1st Sess. 4, reprinted in 1979 U.S.Code Cong. & Ad.News 1469, 1472 (emphasis added).

Such economic coercion will be joined by coercion on litigants from the district courts themselves. It ignores reality to suppose that at least some busy district courts will not control their dockets by pressuring litigants to consent to trial before a magistrate. The steady growth in the use of magistrates has been well documented. From 159 full-time magistrate positions authorized for 1978, the number has risen to 223. See 1977 Annual Report of the Director of the Administrative Office of the United States Courts at 28 (Report); 1983 Report at 200. While the majority and the Third Circuit in Wharton-Thowas, 721 F.2d at 930-931 , recognize that pressure on parties to submit cases increases in direct proportion to the number of magistrate positions, the majority argues that the process can be reversed when the situation becomes intolerable. Majority op. at 546. I cannot agree. The Constitution should prevent this type of coercion from ever occurring.

The Senate Report’s explicit intent to induce the poor to choose magistrates is matched by an equally unsettling expression in a House Report that cases which do not require sophisticated legal knowledge should be given to magistrates, rather than to Article III judges.

H.Rep. No. 1364, 95th Cong., 2d Sess. 12 (1978). Reflecting the philosophy that Article III judges should be reserved for complicated disputes, the district courts have adopted rules referring to magistrates for recommended findings on grievances between individuals and the government, including civil rights cases and habeas corpus. See, e.g., D.Ariz.R. 17(c) (habeas corpus, other post-conviction petitions, prisoner civil rights); C.D.Cal.R. 1.0 (entitlement to Social Security benefits, immigration, civil rights actions under 42 U.S.C. §§ 1981-1986 , habeas corpus, judgment debtor proceedings, and others); D.Col.R. 17(f) (all prisoner petitions).

Yet Hamilton’s vision was that independent judges exercise Article III judicial power to decide all cases in which the operation of our laws is claimed to work an injustice, not just cases that some view as more important:

The Federalist No. 78, at 528 (J. Cooke ed. 1961).

A comparison of that simple statement with the confused allocation of responsibility between Article III judges and magistrates under this Act demonstrates how far we have veered from the course the Constitution charted.

*555 In my opinion, the panel correctly decided that the consensual reference of civil cases to a magistrate violates the Constitution.

. The only cases cited for the proposition that consensual reference to non-Article III decision makers is constitutionally permitted are four nineteenth century Supreme Court decisions. See Kimberly v. Arms, 129 U.S. 512 , 9 S.Ct. 355 , 32 L.Ed. 764 (1889); Newcomb v. Wood, 97 U.S. (7 Otto) 581, 24 L.Ed. 1085 (1878); Heckers v. Fowler, 69 U.S. (2 Wall.) 123 , 17 L.Ed. 759 (1864); Alexandria Canal Co. v. Swann, 46 U.S. (5 How.) 83 , 12 L.Ed. 60 (1847). Not only do these cases contain no separation of powers analysis, they include no mention of Article III whatsoever. Kimberly and New- *b concerned reference to special masters or bankruptcy referees over whom the district court exercised review and directed entry of final judgment. Kimberly, 129 U.S. at 524-25 , 9 S.Ct. at 359-60 ; Newcomb, 97 U.S. at 583 . Heckers and Alexandria Canal involved the common law predecessors to our present arbitration system, which operates outside the Article III forum. See 16 S. Williston, A Treatise on the Law of Contracts § 1919 (3d ed. 1976). It elevates these cases beyond their historical context to argue that they constitute Supreme Court approval for the wholesale delegation of Article III judicial power to non-Article III judges.

. The Magistrates Act also authorizes magistrates to make final decisions in criminal misdemeanor cases. 28 U.S.C. § 636 (a). That provision is not at issue here.