Plaintiff, a dozen years ago, began this suit against the Columbia Broadcasting System, Inc. and allied corporations (CBS) to seek compensation for their unauthorized use of a character concept he had developed, embodying a costume, slogan, name, and symbol. A mechanic living in Cranston, Rhode Island, his avocation had been to don an all black cowboy suit, with a St. Mary’s medal affixed to his flat crowned black hat, a chess symbol to his holster, and an antique derringer secreted under his arm, and make public appearances at rodeos and other events, meeting innumberable children, and passing out his card, inscribed with a chess set knight, proclaiming “Have Gun Will Travel, Wire Paladin, N. Court St., Cranston, R.I.” We described these events in greater detail when this case was before us on appeal from the first count1 in his complaint, Columbia Broadcasting System, Inc. v. DeCosta, (DeCosta I), 377 F.2d 315 (1967), cert. denied, 389 U.S. 1007, 88 S.Ct. 565, 19 L.Ed.2d 603 (1968). As every well-versed television viewer of the late fifties and early sixties knows, the gestalt conveyed by plaintiff’s costume and accessories found its way into defendants’ television series, “Have Gun Will Travel”, which enjoyed enormous popularity for over eight years in its initial run, grossing in excess of fourteen million dollars.
The claim below, in the remaining counts two and three of the complaint, asserted a wilful and intentional infringement of plaintiff’s common law trademark and/or service mark and unfair competition. The plaintiff sought both injunctive and monetary relief, including an accounting for all profits made by defendants in broadcasting “Have Gun Will Travel”. The first count had been tried before a jury. When this court reversed the result obtained in that trial, the case was returned to the district court where the parties entered into a stipulation that counts two and three be determined by the district judge on the basis of the trial transcript, including all exhibits, together with a stipulation of additional testimony. 383 F.Supp. 326, 327.
Subsequent to this stipulation the following order was entered by the district court on October 5, 1973:
“Pursuant to stipulation of the parties on each side and of counsel on each side, the above captioned case is referred, under authority of 28 U.S.C. 636, to United States Magistrate Jacob Hagopian for hearing and determination to be had on or before November 5, 1973.”
Six months later, on April 15, 1974, a report was filed by the magistrate setting forth his findings of fact and conclusions of law sustaining the plaintiff’s position. Five days after this report was filed, the defendants objected to the reference for the first time and argued that the parties were without authority to consent to reference and that the decision of the magistrate had been ultra vires. The district court ruled that the consensual reference granted the magistrate the power to “determine” the issues in the case. It therefore held the magistrate’s exercise of jurisdiction legitimate and restricted its own review to a search for “manifest error” of fact or law. Finding none, it affirmed the magistrate’s decision and entered judgment thereon. This appeal challenges both the propriety of the reference and the decision on the merits.
I. The Reference.
The defendants’ argument to the district court and to us on appeal may be *503reduced to the following syllogism: (1) the order of reference in effect authorized the magistrate to act as special master; (2) the United States Magistrates Act, 28 U.S.C. §§ 631-639 2 provides for appointment of magistrates as masters only under the auspices of the Federal Rules of Civil Procedure; (3) Rule 53(b) provides that “in actions to be tried without a jury, save in matters of account and of difficult computation of damages, a reference shall be made only upon a showing that some exceptional condition requires it”; (4) since there is not such exceptional condition here, the reference was improper and the magistrate was without jurisdiction to hear the case; and (5) since this is a jurisdictional matter, it cannot be waived by counsel. They further argue that even if the consent of the parties ratifies the reference, the district court employed an inappropriate standard of review, and the governing standard should not have been manifest error of fact or law, but errors of law and clearly erroneous findings of fact as provided in Rule 53.
The district court’s analysis began with a holding that a consensual reference to a magistrate was not a reference to a special master within the meaning of 28 U.S.C. § 636(b)(1) and was instead governed by that part of § 636(b) which provides: “any district court may establish rules pursuant to which any full-time United States magistrate may be assigned such additional duties as are not inconsistent with the Constitution and laws of the United States.” Reviewing the law governing the powers of non-Article III3 judges to hear and determine cases, it held that there was no statutory or constitutional bar to the magistrate adjudicating a civil case on the basis of the litigants’ consent. And to determine the scope of review, the district court examined Kimberly v. Arms, 129 U.S. 512, 9 S.Ct. 355, 32 L.Ed. 764 (1889), which treated district court review as limited to the question of “manifest error” in law or fact. 383 F.Supp. at 336-337.
Defendants argue that it is constitutionally impermissible for an Article III judge to abjure decision making responsibilities and that the judge is therefore without power to invest other non-Article III judicial officers or parajudges, such as magistrates, with broad authority. However persuasive such an argument may be where governmental sanction is threatened,4 indicating a strong public interest in the outcome of litigation 5 and creating a countervailing ne*504cessity for extending the full measure of judicial process to the defendant,6 or where parties to civil litigation properly before the federal judiciary insist on judicial resolution,7 quite different policy and precedent should apply where the parties to a civil dispute themselves select another forum. Under such circumstances, it is inappropriate to evaluate the problem as one of the right of the judiciary to relinquish its authority.8 The issue is not the power of the judge to refer, but the power of the parties to agree to another arbiter, absent overriding constitutional considerations.
There is old authority for consensual reference for decision after court proceedings have been instituted. In Heckers v. Fowler, 2 Wall. 123, 69 U.S. 123, 17 L.Ed. 759 (1864), the parties agreed to a reference under the terms of which the report of the referee was to “have the same force and effect as a judgment of the court.” Id. at 127, 17 L.Ed. 759. In responding to a challenge to the power of a federal court to authorize such a proceeding, the Court endorsed the holding of an earlier case that “a trial by arbitrators, appointed by the court, with the consent of both parties, was one of the modes of prosecuting a suit to judgment as well established and as fully warranted by law as a trial by jury . . . .” Id. at 128-129, 17 L.Ed. 759. It also recognized that, assuming the preservation of exceptions to acceptance of the report and a sufficient record, the same scope of appellate review would be available as in the case of appeal from an ordinary trial court judgment. See also Newcomb v. Wood, 97 U.S. 581, 24 L.Ed. 1085 (1878). And, in Kimberly v. Arms, 129 U.S. 512, 9 S.Ct. 355, 32 L.Ed. 764 (1889), the Court said:
*505“A reference by consent of parties, of an entire case for the determination of all its issues, though not strictly a submission of the controversy to arbitration — a proceeding which is governed by special rules — is a submission of the controversy to a tribunal of the parties’ own selection, to be governed in its conduct by the ordinary rules applicable to the administration of justice in tribunals established by law.” Id. at 524, 9 S.Ct. at 359.9
We would add that we find nothing in La Buy v. Howes Leather Co., 352 U.S. 249, 77 S.Ct. 309, 1 L.Ed.2d 290 (1957) to cast doubt on the propriety of truly consensual reference. The Court in La Buy was concerned with the wholesale resort to references to private masters ordered, despite objection, by district courts attempting to deal with crowded dockets, such references having gained notoriety for their delay and high cost. We see none of these problems in truly consensual references to magistrates.
Both Heckers and Kimberly spoke in terms of arbitration supervised by the Court even though both cases long antedated the present availability of authorization for binding private agreement to arbitrate future disputes. In that era, an agreement to arbitrate a dispute that might arise was not judicially enforceable,10 although an arbitration award, once granted, was treated as the equivalent of adjudication by the courts.11 There was a long history in the field of arbitration of unwillingness on the part of the judiciary to relinquish power to decide cases properly before the courts.12 With the adoption of the Federal Arbitration Act, 9 U.S.C. §§ 1 — 14, this judicial hostility toward enforcement of arbitration agreements had gradually changed to approval of this mode of dispute resolution.13 Under section 9 the court must enter judgment upon the award of the arbitrators unless certain abuses are apparent.14 The authority of the courts to relinquish their decision making authority in favor of the arbitrators to the extent described in a private contract has long been assumed to have passed constitutional muster.15
From a constitutional viewpoint, we can see no significant difference between arbitration and consensual reference for decision to magistrates. In both situations the parties have freely and knowingly agreed to waive their access to an Article III judge in the first instance. Or put another way, they have chosen another forum. Indeed, the decision to waive in the case of a consensual reference is more knowledgeable than in the case of an agreement to arbitrate a future dispute because it is made after the issue has crystallized. Both modes of conflict resolution serve the same goals of relieving scarce judicial resources and of accommodating the parties. If it be *506queried whether the dignity of Article III is being compromised by entering judgments on awards made by non-Article III personnel, the sufficient rejoinder is that judgments are entered on arbitrators’ awards.
Defendants, however, urge that, even if there is no constitutional defect vitiating consensual references to magistrates, such are now proscribed by the Magistrates Act in light of its legislative history. It is true that the legislative history of the Act clearly demonstrates an intent to subject the court’s power to compel a reference of ordinary cases to the limitations of the Federal Rules (including Rule 53(b), preconditioning reference on a finding of exceptional conditions). That question, however, is not now before us.
The question here is whether the legislative history indicates that the same strictures were to be placed on consensual reference. Very little was said on this subject. Most of the voluminous testimony and written presentations were concerned with the duties which, as of course, could be assigned to magistrates. But what little was said makes it clear that it was assumed as given that consensual references were available. Defendants have not seen fit to acknowledge the relevant legislative history.
In the first place, the draft of the legislation which was to become the Magistrates Act was accompanied, as early as April of 1966, by a Senate subcommittee staff memorandum, quoted by the district court, stating: “The use of magistrates for duties that do not require the employment of an Article III judge, or in cases in which the parties consent to the use of a magistrate, may do much to increase the efficiency of the Federal Courts.” Hearing on S. 3475 before the Subcom. on Improvements in Judicial Machinery of the Senate Comm, on the Judiciary, 89th Cong., 2d Sess., at 14 (1966).
A second indication of Congressional intent stems from the subcommittee’s concern over the challenge made by then Assistant Attorney General Fred M. Vinson, Jr. to the proposed assignment of magistrates to try minor offenses if the defendant knowingly and intelligently consented.16 The subcommittee then charged its staff to make a study, which resulted in a memorandum addressed to “The Constitutionality of Trial of Minor Offenses' by U. S. Magistrates”. Hearings on S. 945 before the Subcom. on Improvements in Judicial Machinery of the Senate Comm, on the Judiciary, 90th Cong., at 246 — 256 (1967). The memorandum concluded that the proposed minor offense provision of the Magistrates bill was constitutional, drawing on factors supported by three separate lines of authority: (1) the fact that the magistrate would be an officer of the court and subject to its control; (2) the necessity of consent [“If any ‘right’ to trial by a good-behavior tenure judge exists, it is a due process right, which is waivable like other due process rights.”]; and (3) the preservation of a right of appeal to the court. Id. at p. 254.17 In the course of the memorandum, a separate subsection entitled “Consensual reference to special master” noted that “In ordinary practice before the Federal district courts, the parties can also consent to delegation of judicial responsibility to a court officer — -a special master.” and quoted from Kimberly v. Arms, supra. Id. at p. 253. 18
*507Finally, in the Senate Report presenting the legislation to the Senate, S.Rep.No. 371, 90th Cong., 1st Sess. (1967), the following passage,- quoted by defendants, appears at p. 25:
“(1) A magistrate may be assigned to serve as a special master in an appropriate civil action, under the conditions imposed by pertinent provisions of Federal Rules of Civil Procedure. This latter requirement was added as a clarification in response to comments made by the Committee on the Administration of the Criminal Law of the Judicial Conference. In particular, rule 53(b) provides that a reference to a master shall be the exception and not the rule; that in actions tried by a jury a reference shall be made only when the issues are complicated; and that in actions tried without a jury, save in matters of account, a reference shall be made only upon a showing that exceptional circumstances require it. These conditions, which in essence reflect the rule laid down by the Supreme Court in La Buy v. Howes Leather Company, 352 U.S. 249 [77 S.Ct. 309, 1 L.Ed.2d 290] (1957), protect against any abdication of the decision-making responsibility that is properly that of the district courts.”
Defendants omitted the subsequent sentence, which concluded the paragraph:
“A memorandum entitled ‘Restrictions on the Use of Special Masters,’ prepared by the staff of the Subcommittee on Improvements in Judicial Machinery and published at pages 281 — 283 of the hearings before that subcommittee on the Federal Magistrates Act, provides an analysis of rules relevant to the assignment of magistrates as masters.”
Recourse to the memorandum referred to reveals a paragraph entitled “General Restrictions”, which ends as follows:
“Furthermore, a reference cannot be made to a master if it would result in an infringement of a party’s right to trial by judge and/or jury or on any other substantive right. Schwimmer v. United States, 232 F.2d 855 (8th Cir.), cert. denied, 352 U.S. 833, 77 S.Ct. 48, 1 L.Ed.2d 52 (1956). But a party can waive these rights by consenting to a reference. Allen Bradley Co. v. Local 3, IBEW, 51 F.Supp. 36 (S.D.N.Y.), reversed on other grounds, 145 F.2d 215 (2d Cir.), which was reversed, 325 U.S. 797, 65 S.Ct. 1533, 89 L.Ed. 1939 (1943); see Hart v. Williams, 91 U.S.App.D.C. 340, 202 F.2d 190 (1952).” Hearings on S. 945, supra at 282.
We see no contrary references in the legislative history. It therefore seems clear to us that the Congressional intent was to leave untouched the tradition, as Congress understood it, that parties could, without violation of Article III freely consent to refer cases to non-Article III officials for decision. We see no suggestion in the committee reports and no reason of policy which would limit consensual reference to “some exceptional conditions” as required by F.R.Civ. Pro. 53(b). As we have indicated above, the factors triggering the decision in La Buy are not present when the parties knowingly and freely agree to a reference to a magistrate. Indeed, on occasion, perhaps when the legal issue is closely balanced and the stakes are not high, or when expedition and expense are dominating factors, parties may prefer prompt decision, through by a magistrate, to decision by a judge. This avenue ought not to be barred. But in order that such references not be subtly coerced by a harried court, intentionally or not, district courts should adopt a procedure such as requiring that the parties file with the clerk of the court a letter of consent to have the magistrate render decision in the case, the clerk being bound not to disclose the identity of any who consent or who withhold consent.
That it is constitutionally and statutorily permissible to refer cases, *508with the consent of all parties, for initial decision is clear. What does not appear at all clear to us is whether the decision of the referee once rendered is by statute subject to very limited review or to review for any prejudicial legal error as required by F.R.Civ.Pro. 53(e)(4). Mr. Justice Black observed in Commonwealth Coating Corp. v. Casualty Co., 393 U.S. 145, 149, 89 S.Ct. 337, 339, 21 L.Ed.2d 301, (1968), in a case involving an arbitrator charged with partiality, “[W]e should, if anything, be even more scrupulous to safeguard the impartiality of arbitrators than judges, since the former have completely free rein to decide the law as well as the fact and are not subject to appellate review.” If review of arbitration awards can be limited to issues involving the integrity of the process (not the correctness of the decision), it may well be that review could be similarly limited as to consensual references for decision. But the actual scope of review that has been provided in the past is unclear. Various references we have cited suggest that the report of a referee to whom a case has been consensually referred for decision stands in the same position as a district court decision, reviewable by a court of appeals for clearly erroneous factual findings and errors of law. This would avoid an added tier of decision, but at the expense of losing the benefit of the district court’s deliberation. The “manifest error” language in Kimberly would indicate a reduced standard, which the district court applied in this case. We have indicated our feeling that as applied to legal error, the term lacks content. It is possible that, following the analogy of arbitration, a truncated review to insure integrity of the process could be devised. But in the present state of the law we would be reluctant to approve even a clearly worded consensual reference to a magistrate which purports to finally bind the parties to his rulings of law. Until Congress, on reviewing the experience under the Federal Magistrates Act, fashions a review procedure for consensual reference for final (or semi-final) determination of all issues of law and fact, it might be better to rely on the formulation contained in Rule 53(e)(4).19
We need attempt no categoric answer to this issue since the reference in this case was not clear enough by its own terms to support the conclusion that the parties consented to a grant of power to the magistrate greater than that outlined in Rule 53 Fed.R.Civ.Pro.20 The district court relied on the language of the order indicating that the parties agreed to “hearing and determination”. But this language is fully compatible with the magistrate as trier of fact whose rulings on the facts are final under Rule 53(e)(4), but whose legal rulings have no binding force. We agree with the district court that the defendants’ motion for summary judgment is some evidence of an intent to have the magistrate rule on issues of law, but, in light of the absence of any specific court rule addressing itself to the effect of a magistrate’s decision and the failure to include in the order of reference a provision for entry of judgment upon the magistrate’s award, we do not find this sufficient.
We must, therefore, treat the district court’s failure to review the magistrate’s decision on the law under Rule 53 *509as error. It may well be noted that, deciding that a different standard of review should be applied, logic should compel us to remand to the district court. But we would have to repeat the legal review and we see no favor to the parties in prolonging further these aging proceedings. We proceed, then, treating the factual findings of the magistrate adopted by the district court as subject to the clearly erroneous standard of Rule 52(a), and treating the legal decision of the magistrate as subject to the full review of this court.
II. The Merits.
We proceed to review the magistrate’s findings in the light of the pleadings, the evidence, and the law. In DeCosta I we considered only the first cause of action in the complaint — that for misappropriation of “the idea and character”, seeking damages based on defendants’ unjust enrichment. Before us now are plaintiff’s second and third causes of action, for trademark and/or service mark infringement and for unfair competition. In the second cause of action, plaintiff alleges that the slogan “Have Gun Will Travel”, together with a figure of a knight chess piece and the words “Wire Paladin”, as used in his calling cards, constituted his distinctive common law service mark or marks, which defendants intentionally infringed, appropriating the goodwill created by plaintiff, causing him serious financial damage. In the third cause of action plaintiff alleges that defendants intentionally copied plaintiff’s service marks and his manner of dress (black outfit, white tie, mustache, etc.), .and passed off their television character of Paladin as the original. Paladin portrayed by plaintiff, profiting and diluting plaintiff’s established goodwill, thereby deriving unjust profits. Plaintiff sought injunctive relief, an accounting for all profits, a trebling of “profits lost and damages suffered by plaintiff”, plus a separate item of $2,000,000 damages.
The magistrate’s well ordered opinion on cross motions or summary judgment was divided into findings of fact, a discussion of preemption and other threshold legal issues, and conclusions of law. After rehearsing the general factual background described in DeCosta I, 377 F.2d at 316-317, he made the following critical findings: (1) As to secondary meaning — some 21 witnesses knew plaintiff as Paladin, 18 had seen him pass out his cards and plaintiff received cards and letters addressed to him as Paladin; (2) As to confusion — the card and costume marks and dress of plaintiff were identical to those used in defendants’ TV series, with attendant “great likelihood of confusion”; 21 witnesses discerned a similarity between plaintiff and the television Paladin, six first thought the television Paladin was plaintiff; (3) As to plaintiff’s commercial activity — he “commercially held himself out to be ‘Paladin’ ” and used his marks in interstate commerce to advertise his services as an entertainer, and in connection with his appearances at his pony ring where children paid for rides, and appearances at paid admission rodeos.
In his legal discussion the magistrate held that DeCosta I had not precluded recovery on the second and third causes of action; that the Sears-Compco decisions 21 had not preempted application of the common law of service marks and unfair competition or of the Lanham Act, 15 U.S.C. §§ 1125 and 1127; that a belated res judicata claim of defendants based on a trade mark administrative proceeding was not available; that plaintiff was not barred by laches in bringing this action six years after the alleged infringement; and that New York policy and law (as well as the Lanham Act, which is consistent therewith) applied.
Finally, the magistrate concluded that plaintiff’s marks were not within the scope of copyright laws; that they had been used in interstate commerce, and had acquired a secondary meaning; and that, although^ there' was no direct competition involved defendants had infringed, had falsely advertised them to be their own, and had therefore unfairly *510competed. He held that defendants should account to plaintiff for all profits arising from the use of plaintiff’s service marks.
We note at the outset of our analysis that two allegations of the complaint meet with no response in the findings of the magistrate; in the second cause of action the allegation of financial damage, and in the third cause of action, the allegation of “passing off”. Since plaintiff’s role-playing as Paladin was uncompensated, we think that any finding of financial damage would have been unsupported. We say the same about “passing off”; there is no evidence that the CBS nationwide television venture ever sought to draw luster from the Cranston paradigm. We recognize that precision of pleading no longer can be an absolute determinant of either liability or remedy. What, if any, effect these gaps between pleading and proof have we leave for later analysis.
Picking our way through the cluster of issues in several related but different fields of the law, we start where we left off, at DeCosta I. We there gave what has since been characterized as an expansive reading to Sears and Compco. See, e. g., Comment, Copyright Pre-emption and Character Values; The Paladin Case as an Extension of Sears and Compco, 66 Mich.L.Rev. 1018 (1967— 1968). We held that “if a ‘writing’ is. within the scope of the constitutional clause, and Congress has not protected it, whether deliberately or by unexplained omission, it can be freely copied.” 377 F.2d at 319. Since the cards— including the photograph — “were unquestionably ‘writings’ within the meaning of the copyright clause, and arguably were copyrightable under the statute”, id. at 321, we concluded that plaintiff could prevent others from copying them.22
We now know, after Goldstein v. California, 412 U.S. 546, 560, 93 S.Ct. 2303, 2311, 37 L.Ed.2d 163 (1973), that “under the Constitution, the States have not relinquished all power to grant to authors ‘the exclusive Right to their respective Writings’.” In Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470, 478-479, 94 S.Ct. 1879, 1885, 40 L.Ed.2d 315 (1974), this holding was summarized: “at least in the case of writings, the States were not prohibited from encouraging and protecting the efforts of those within their borders by appropriate legislation.” Were we to take this literally, we would see no reason to look on Goldstein as relevant to this case; plaintiff relies on a common law service mark as subsumed also by a federal statute, 15 U.S.C. § 1125 (part of the Lanham Act), but not state legislation. Nevertheless, since the countervailing state law found sufficient in Kewanee was common law, not statutory law, we do not read its capsule of Goldstein literally.
We face a dilemma. Goldstein tells us that we were, in our interpretation of the preemptive reach of the Copyright Clause, over-inclusive. And yet, what we decided in DeCosta I has settled, for this case, the issue of misappropriation. Had Sears and Compco remained unglossed, we might well rule that DeCosta I had predetermined our decision on the second and third causes of action. For if defendants could not be prohibited from copying the cards, which gave the defendants all that they took from plaintiff and incorporated in their series, would not any sanction, whether based on service mark infringement or unfair competition, denigrate their license? We do not, however, rest on any such implications of De Costa I, realizing that we do not have a tabula rasa to write upon.
A starting point, of course, is the explicit domain left open in Sears, that a *511state “may protect businesses in the use of their trademarks, labels, or distinctive dress in the packaging of goods so as to prevent others, by imitating such markings, from misleading purchasers as to the source of such goods.” 376 U.S. 225, 232, 84 S.Ct. 784, 789.23 A second guideline is that in DeCosta I we dealt only with copying “simpliciter”; to the extent that other elements of other causes of action are relevant, we are free to consider them. In the context of the instant case, we are free to consider deception and passing off, secondary meaning and likelihood of confusion.
Our first inquiry is whether plaintiff’s marks and dress are protectible under either the broad concept of unfair competition or that part of unfair competition dealing with trade and service mark infringement. As to plaintiff’s dress, this may have had less claim to protection against infringement than his card. For there is no question but that the photograph of plaintiff in full regalia which he widely distributed was copyrightable. 17 U.S.C. § 5(j), Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 4 S.Ct. 279, 28 L.Ed. 349 (1884). The card is less vulnerable on copyright preemption grounds. Its three components — the slogan, “Have Gun Will Travel”, the name in the phrase “Wire Paladin”, and the knight chess piece — are in categories protectible as common law trade or service marks. See, e. g., Call-man, The Law of Trademarks, Vol. 3, § 77.4(g) (slogans); § 66.1 (trade names).
What is not clear is whether plaintiff’s activities in connection with his marks entitle him to the protection of trade and service mark law. As we said in DeCosta I, 377 F.2d at 316, “[t]his was perhaps one of the purest promotions ever staged, for plaintiff did not seek anything but the entertainment of others.” Plaintiff testified that he never used the name “Paladin”, the slogan or the chess piece for any business use, and never published any advertisements; that he did not even receive his expenses for appearances at horse shows and rodeos; and that the distribution and mailing of pictures and cards was costly to him. The closest point to commercialism that plaintiff came was in conducting a horse trading business at a ranch, where he stapled some of his cards to some beams near the ceiling of a barn. In 1953, he left that ranch and established a pony ranch, sold rides and, in costume, passed out cards and photographs to children. This business ended in September of 1957, when plaintiff had to move out. Ironically, this was the month when defendant’s television series commenced. From then through 1964 plaintiff continued to appear in parades, at hospitals, and at other gatherings, but was not, apart from his employment as a mechanic, associated with any business venture.
While the magistrate found that plaintiff “commercially held himself out” to be “Paladin”, and that he used his service marks in interstate commerce to advertise his services, we interpret his holdings not as purely factual ones but as decisions combining both law and fact. That is, the findings could not, on this record, be that plaintiff was engaged in a remunerative business, sparked by the profit motive. Rather, they were findings that plaintiff regularly (on week-ends) held himself out as a personality and entertainer, appearing in various states, and that, even though he incurred expense without receiving income, his activity constituted the rendering of service in interstate commerce. The Lanham Act, 15 U.S.C. § 1127, is, for our purposes, delphic, that is, circular in its relevant definitions. It defines a service mark as “a mark used in the sale or advertising of services to identify the services of one person and distinguish them from the services of others.”24 “Commerce” means “all commerce which may lawfully be regulated by Congress” and a service mark is deemed to be used *512in commerce “when it is used or displayed in the sale or advertising of services and the services are rendered in commerce, or ... in more than one State . . . and the person rendering the services is engaged in commerce in connection therewith.”
Defendants rely on cases such as Hanover Star Milling Co. v. Metcalf, 240 U.S. 403, 36 S.Ct. 357, 60 L.Ed. 713 (1916) and United Drug Co. v. Rectanus Co., 248 U.S. 90, 39 S.Ct. 48, 63 L.Ed. 141 (1918), for support of the proposition that a trade mark is inseparable from the goodwill of a business. Defendants also cite the recent case, Heinemann v. General Motors Corp., 342 F.Supp. 203 (N.D.Ill.1972) , aff’d, 478 F.2d 1405 (7th Cir. 1973) . In that case a plaintiff had “modified” an ancient Model A Ford and had named it “The Judge”, after the well known skit in “Laugh In”, a television show. He raced and exhibited his car, occasionally receiving a mileage allowance, radio advertising, and free entry in a race. He intended to open an automobile equipment shop and capitalize on the good will generated by the slogan. After defendant General Motors brought out a new Pontiac called “The Judge”, plaintiff adopted another name. The court rejected plaintiff’s claims for trade and service mark infringement and unfair competition, holding, citing only Rectanus, supra, that the slogan, to be profitable, must be used in connection with a trade or business at the time of the infringing activity. All plaintiff had was the desire to open a business in the future, together with the hobby of racing and displaying his automobile.
From a policy point of view, the functions served by a trademark (or service mark) — an indication of origin, a guarantee of quality, and a medium of advertisement, 3 Callman, § 66.3, p. 36— suggest association with enterprises selling their goods or services. Callman refers to a service mark as “the mark of a business that only renders service and is not engaged in the manufacture or sale of goods.” Id. § 68.1, p. 69. And, again, “The service mark must, of course, be used as an identification mark, and it must perform the same functions in the selling and advertising of services that a trademark traditionally performs in respect of goods.” Id. § 68.1, p. 71.25 Our impression, derived from our scrutiny of the cases, the treatises with their myriads of citations, and the periodical literature including the Trade Mark Reporter, is that virtually all the litigation and doctrine center about the pursuit of economic gain.
We say “virtually” because the law of unfair competition now protects “eleemosynary organizations [which] function in commerce and, in form, resemble business enterprises . . The happenstance that they are nonprofit-seeking ventures, and therefore removed, as such, from the rigors of business competition, neither eliminates the element of competition nor disentitles them to protection against the unfair competition of similar organizations.” Callman, Vol. 1, § 1.1, p. 4. See, e. g., The President and Trustees of Colby College v. Colby College — New Hampshire, 508 F.2d 804 (1st Cir. 1975). While the quoted passage speaks only of protection against unfair competition from similar institutions, we have no doubt that a non-profit institution may protect its name against filching by a commercial company.' It is not a big step in theory to say that an individual who develops and promotes for entertainment purposes a specialty character associated with a distinctive name, costume, slogan, and other marks, but who never charges a fee, is also entitled to protection under the common law doctrine relating to trademarks and unfair competition. But, if the step were taken, what principle should both guide and limit? If plaintiff is recognized as one who performs services in commerce, what about the hobbyist magician, square dance caller, story teller, amateur actor, singer, barbershop quartet, stand*513up comic? How frequent must be the appearances? How prominent the mark? How often must the performer appear in other states to enjoy Lanham Act protection? We know of no logical or obvious line of demarcation. The price of an open-ended extension of protection for some creative activity is, pro tanto, a curtailment of the borrowing, poaching, imitation which underlie so much other innovative activity.
Protection at present has the merits of inherent limitations: the existence of a trade, business, or profession where the “good will” to be protected has been subject to the acid test of the willingness of people to pay for goods or services; or, in the case of nonprofit institutions, the voluntary investment in time, effort, and money of many individuals to create and maintain a program of sufficient interest to consumers, members, and sponsors to warrant protection. We hesitate to take the step of offering common law unfair competition protection to eleemosynary individuals. Whether legislatures are better equipped than courts to deal with this problem, we cannot clearly say, but in our posture of doubt would prefer to see expansion of protection come from that source.
Having exposed our misgivings, we do not rely on a holding that the absence of a profit-oriented enterprise disqualifies plaintiff from protection. We shall assume, therefore, that plaintiff’s marks meet the requirements of common law service marks. We also shall assume that they are distinctive enough so that proof of secondary meaning is not essential. In the alternative, we shall accept the finding of the magistrate, adopted by the district court, that, at least among some people, plaintiff’s name and card had come to be associated with him.
This brings us to the critical issue: whether there was a deceiving of the public as the result of defendants’ actions. As we have noted, there was not and could not supportably have a been a finding that defendants passed off or palmed off their “Paladin” and television program as the creations of plaintiff. Having already decided in De-Costa I that there was no culpable misappropriation, we are left with the issue of confusion, or its likelihood.26 As Call-man has said, “It is the gravamen of an action for trademark infringement that the defendant’s use of a trademark similar to the plaintiffs’ created a likelihood of confusion.” Callman, Vol. 3, § 80, p. 538. He states further that “There is a confusion of goods when an ‘ordinarily prudent purchaser would be liable to purchase one product in the belief that he was purchasing the other,’ ” Id. at 541 [footnote omitted]. In resolving this issue there are several guidelines: there need not be a similarity of competing activities, id. at 539, or goods or services of the same description; confusion may consist in misleading as to the identity of goods, or of source, or of a relationship to the originator, id. at 543 — 546; and “Mere visual or side-by-side comparisons of the trademarks will not alone suffice. Extrinsic facts are highly significant, as, for example the types of goods, the classes of prospective purchasers, the competitive relationship of the parties, third party registrations, the extent of the public’s awareness of the plaintiff’s trademark, the manner of trademark use (visual, auditory, or mental), the degree of the defendant’s necessity to use his mark, and the trend towards expansion in the respective trade or industry as to territory or line of merchandise.” 27 Id. at 556.
*514There is another factor, ¡we think, which should guide a judgment as to likelihood of confusion — the time when the evidence is submitted. There is ample reason, at the incipiency of an alleged infringement, in a suit seeking injunctive relief, for a plaintiff to argue and a court to rule that the similarity of marks is such that confusion is all too likely to ensue. Plaintiff should not be expected to stand by and await the dismal proof. Callman, Vol. 3, § 80.6, pp. 559 — 560. “[C]onversely”, adds Callman, “after the lapse of substantial time if no one appears to have been actually deceived that fact is strongly probative of the defense that there is no likelihood of deception arising out of the use of the mark in question.” Id. at 562. Defendants have urged that plaintiff, having delayed for almost the entire limitations period before bringing suit, should be barred by the doctrine of estoppel by laches from further prosecution of this litigation. Plaintiff’s rejoiner is that he was, during this period, going from attorney to attorney, seeking assistance before he was able to make progress. We do not find in the authorities sufficient support, where defendants’ action was found by the jury to be deliberate and knowing, to invoke this doctrine. Nevertheless, we do say that the delay, for reasons stated, increases the quantum of proof of confusion which the plaintiff has the obligation to supply.
When we examine the magistrate’s opinion on this issue, we find only a finding as to ■ the identical nature of the marks used by plaintiff and defendants and the resulting “great likelihood of confusion” therefrom, and another that “At least 6 . witnesses testified that they had at first thought, on viewing the television program, that Richard Boone was the plaintiff Mr. De-Costa until they learned the contrary from viewing the credits of the show, or otherwise.” 28 The paucity of these findings takes on some significance when we note that after remand of the case to the district court in 1968, following our decision in DeCosta I, the plaintiff was given the opportunity to supplement the record in support of his second and third causes of action. The quoted findings are the result.
The magistrate’s findings being treated as final, they are here subject to the same standard of review as are district court findings under Fed.R. Civ.Pro. 52(a). They must be accepted unless clearly erroneous. On this issue, however, we see no alternative to saying the finding is not supported. If the identity of the marks settled the question of likelihood of confusion, there was no need for further evidence after De-Costa I; we had taken some pains to make the point. But, in fact (rather, in law) extrinsic evidence, as we have noted, should be considered. Here, the disparities are substantial. Plaintiff’s enterprise was localized; defendants’ was nationwide. Plaintiff’s appearances were simple happenings, passing out cards and pictures, riding, occasionally walking his horse, and executing a quick draw gun maneuver; defendants’ extrinsic television series portrayed an elegant hired gun on manifold missions. Plaintiff’s “customers” were attendees at rodeos and parades, patients at hospitals, etc.; defendants’ purchasers were its program’s sponsors, who in turn were responsive to the nation’s television audience. Moreoever, plaintiff’s suit was not brought in 1957 when the alleged infringement began; the testimony as to likelihood of confusion was introduced by affidavit eleven years later, when the first run of defendants’ series had terminated. After this lapse of time, the testimony of six witnesses that they thought, on first viewing the program, that the television character Paladin was plaintiff, seems to us either no evidence at all or such minimal evidence as not to *515support a finding of likelihood of confusion requiring an accounting of defendants’ profits from a highly successful, 225 episode series grossing over fourteen million dollars. We do not blame plaintiff or counsel; we suspect that the most exacting search for proof would not have produced more.
We recognize that plaintiff has lost something of value to him. The very success of defendants’ series staturated the public consciousness and in time diluted the attractiveness of plaintiff’s creation, although he continued his appearances longer than defendants’ first run. While he was not injured financially, there can be no doubt that he has felt deprived. As a commentator has observed, “[I]t could be argued that the most appropriate measure of damages would be the emotional harm that he suffered when CBS exploited his character and lured his audience away.” 66 Mich.L.Rev. at 1034. But to give any relief, however tailored, we need a predicate of liability. Absent the ultimate fact of confusion, we cannot find a basis for liability for common law service mark infringement or unfair competition.
Judgment reversed. Remanded with instructions to enter judgment for defendants.
. The first count charged a wilful misappropriation through copying plaintiffs idea and character.
. 28 U.S.C. § 636(b) provides:
“Any district court of the United States may establish rules pursuant to which any . . . United States magistrate, . . may be assigned such additional duties as are not inconsistent with the Constitution and laws of the United States. The additional duties authorized by rule may include, but are not restricted to— “(1) service as a special master in an appropriate civil action, pursuant to the applicable provisions of this title and the Federal Rules of Civil Procedure for the United States district courts;
“(2) assistance to a district judge in the conduct of pretrial discovery proceedings in civil or criminal actions; and “(3) preliminary review of application for posttrial relief made by individuals convicted of criminal offenses, and submission of a report and recommendations to facilitate the decision of the district judge having jurisdiction over the case as to whether there should be a hearing.”
. U.S.Const. Art. III. A magistrate is an Article I judge, being a “tribunals inferior to the Supreme Court” and appointed pursuant to Congressional power under U.S.Const. Article I, Section 8, clause 9. See Glidden Co. v. Zdanok, 370 U.S. 530, 543, 552, 82 S.Ct. 1459, 8 L.Ed.2d 671 (1962).
. Notwithstanding such an argument, the Congress did extend to magistrates the authority to try minor offenses when all parties consent.
. For example, in arbitration cases public policy considerations and Congressional intent have an impact on the available forum. See, e. g„ Wilko v. Swan, 346 U.S. 427, 74 S.Ct. 182, 98 L.Ed. 168 (1953) (the non-waiver provision in the Securities Act forbids bargaining away any right under that Act, including the choice of forum); A&E Plastik Pak Co. v. Monsanto, 396 F.2d 710 (9th Cir. 1968) (the validity of a contract under the Sherman Antitrust Act is not an arbitrable issue); American Safety Equip Corp. v. J. P. Maguire & Co., 391 F.2d 821 (2d Cir. 1968) (a claim under the *504antitrust law states a claim in the public interest which Congress intended the courts to resolve). Similarly in Wingo v. Wedding, 418 U.S. 461, 94 S.Ct. 2842, 41 L.Ed.2d 879 (1974), the Supreme Court limited its inquiry as to the propriety of a magistrate holding evidentiary hearings in habeas corpus cases to the question of whether the demonstrated Congressional intent of the Habeas Corpus Act, 28 U.S.C. § 2243 and the Magistrate’s Act, 28 U.S.C. § 636(b)(3) precluded such a result.
. See Crowell v. Benson, 285 U.S. 22, 51, 52 S.Ct. 285, 76 L.Ed. 598 (1932); see also discussion of applicability of the doctrine to judicial reference in Note, Masters and Magistrates in the Federal Courts, 88 Harv.L.Rev. 779, 787-789 (1975).
. In the pre-Magistrates Act case of La Buy v. Howes Leather Co., 352 U.S. 249, 77 S.Ct. 309, 1 L.Ed.2d 290 (1957), the district court had referred the private antitrust suit sua sponce to a master for hearing and preparation of findings after the judge had heard numerous preliminary motions and had become familiar; with the case. All parties objected to the reference. The Court does not appear to rest on a constitutional analysis, see 88 Harv.L.Rev. at 794, but rather on its supervisory powers, and the known disadvantages of referral to non-official masters paid by the parties. La Buy at 253 n. 5, 77 S.Ct. 309. See also Good Eating Places v. Best Places to Eat, Inc., 131 F.2d 809 (7th Cir. 1942).
In TPO, Inc. v. McMillen, 460 F.2d 348 (7th Cir. 1972), the defendant had pressed for an early trial date, the judge sua sponte ordered reference to a magistrate of a motion to dismiss. Objection was made. After an exhaustive study of the legislative history of the Magistrate’s Act, the court concluded that in the absence of consent, Congress did not intend to grant magistrates decision making power in civil cases. 460 F.2d at 359-361 and see nn. 63, 72. Neither the La Buy court nor the TPO court considered the impact of the antitrust policy upon the cases both brought thereunder. See n. 5 supra.
In Wilver v. Fisher, 387 F.2d 66, 69 (10th Cir. 1967), the parties did agree to reference to a master of supervision over interrogatories and the court was critical of the undertaking of such a proceeding by a master, but the Magistrates Act § 636(b)(2) now specifically provides for this kind of assistance to the district judge in the conduct of pre-trial discovery proceedings.
. Our own prior cases on magistrates deal with such situations and are not apposite here. O’Shea v. United States, 491 F.2d 774 (1st Cir. 1974), was a habeas corpus case which allowed more latitude to the magistrate than was subsequently approved in Wingo v. Wedding, 418 U.S. 461, 94 S.Ct. 2842, 41 L.Ed.2d 879 (1974). In Rainha v. Cassidy, 454 F.2d 207 (1st Cir. 1972), this court faced another habeas corpus case. And in Reed v. Board of Election Comm’rs of City of Cambridge, 459 F.2d 121 (1st Cir. 1972), there was no consent or understanding by the parties as to the scope of the reference, and we disapproved the procedure as we had intimated we would do in Devcon Corp. v. Woodhill Chemical Sales Corp., 455 F.2d 830 (1972).
. Despite its reference to “the ordinary rules applicable ... in tribunals established by law”, the Court, in the succeeding sentence, said that review would lie to correct “manifest error in the consideration given to the evidence, or in the application of the law, but not otherwise.” Id This not only seems to us an internal inconsistency but introduces us to a concept currently without content.
. W. Sturges, Commercial Arbitration and Award, § 15 (1930).
. Id at § 235 at 613; S. Williston, Law of Contracts § 1919 (3d ed. 1938): Simpson, Specific Performance of Arbitration Contracts, 83 U.Pa.L.Rev. 160, 163-164 (1934).
. See Kulukundis Shipping v. Armtorg Trading Corp. (Frank, J.), 126 F.2d 978 (2d Cir. 1942).
. See Note, The Consequences of a Broad Arbitration Clause Under the Federal Arbitration Act, 52 B.U.L.Rev. 571 (1972).
. 9 U.S.C. § 9 provides that judgment “shall be entered upon the award . . . unless the award is vacated, modified, or corrected as prescribed in sections 10 and 11.” Under these latter sections the court’s authority to vacate an award is limited to situations of fraud, bias, or misconduct and its power to modify or correct is limited to material miscalculations or assumption of jurisdiction not granted to the arbitrator.
. See Prima Paint Corp. v. Flood & Conklin Mfg. Co., 388 U.S. 395, 87 S.Ct. 1801 (1967); Bernhardt v. Polygraphic Co. of America, 350 U.S. 198, 76 S.Ct. 273, 100 L.Ed. 199 (1956); see also United Steelworkers of America v. Warrior & Gulf Navigation Co., 363 U.S. 574, 578, 80 S.Ct. 1347, 4 L.Ed.2d 1409 (1960).
. Mr. Vinson’s testimony and prepared statement are at pages 107-131 of these Hearings. Defendants cite the point made by this witness as throwing doubt on the capacity of parties to waive trial before an Article III judge in all cases. They do not refer to the subsequent developments referred to in the text. See also n. 4 supra.
. We do not pass on the validity of these perceptions as applied to criminal cases today, but employ them only to establish the Congressional perspective at the time of enactment.
. It may be of some significance that the Committee in quoting the language we have alluded to in n. 9, supra, saw fit to italicize the words “but not otherwise”. While it thus called attention to the “manifest error” language of Kimberly, it also indicated that it understood the law to be that while a special *507master’s report under a compelled reference is advisory only, “it is just as conclusive as the decision of the court when the litigants consent to the reference.” Id. at 253.
. Rule 53(e)(4) reads as follows:
“Stipulation as to Findings. The effect of a master’s report is the same whether or not the parties have consented to the reference; but, when the parties stipulate that a master’s findings of fact shall be final, only questions of law arising upon the report shall thereafter be considered.”
. The district court itself recognized that the wording of the order was ambiguous, 383 F.Supp. at 331, n. 9, but rejected any constitutional challenge premised on that lack of precision, citing Yascavage v. Weinberger, 379 F.Supp. 1297, 1303 (M.D.Pa.1974). We think that case is not authority for the one now before us. Where the scope of the arbiter’s authority is to be determined by a voluntary undertaking, the document must clearly demonstrate the terms of the referral.
. Sears Roebuck & Co. v. Stiffel Co., 376 U.S. 225, 84 S.Ct. 784, 11 L.Ed.2d 661 (1964); Compco Corp. v. Day-Brite Lighting, Inc., 376 U.S. 234, 84 S.Ct. 779, 11 L.Ed.2d 669 (1964).
. To the extent that a “character” was “ineffable”, incapable of “concrete describable manifestation of intellectual creation”, and therefore ineligible for copyright protection, we thought it also ineligible for any protection, state or federal. Pre-Goldstein, this supposition had respectable support. See Nimmer, The Law of Copyright, Vol. 1, § 30, p. 137, n. 595.
. While the Court makes reference to goods, we construe the caveat to apply also to common law service marks.
. It adds, “Titles, character names and other distinctive features of radio or television programs may be registered as service marks notwithstanding that they, or the programs, may advertise the goods of the sponsor.”
. See, also, Callman, Vol. 1, § 2.2 for a discussion of good will, in which the common core, whatever may be the generating cause, is economic and pecuniary advantage of a going concern.
. This is consistent with the approach of the magistrate, who held that New York law was applicable. He cited and quoted, inter alia, Geisel v. Poynter Products, Inc., 238 F.Supp. 261 (S.D.N.Y.1968), which, in applying New York law, said, “The plaintiff is not required to prove actual palming-off. A showing of the likelihood of customer confusion as to the source of the goods is sufficient.”
. Following the passage, Callman adds, in his 1974 Cumulative Supplement, p. 46, “Such extrinsic facts may be so controlling that, notwithstanding the identity of the conflicting trademarks, there is no likelihood of confusion.”
. Another finding, that 21 witnesses thought on viewing the television program that “There was a physical similarity” between plaintiff and the television character falls short of proof that they were “actually deceived”.