This is the second appeal in this case. Our first opinion is reported in Anti-Monopoly, Inc. v. General Mills Fun Group, 9 Cir., 1979, 611 F.2d 296 (Anti-Monopoly I). On remand the district court again found that the “Monopoly” trademark was valid and had been infringed by Anti-Monopoly, Inc. Anti-Monopoly, Inc. v. General Mills Fun Group, Inc., N.D.Cal., 1981, 515 F.Supp. 448 (Anti-Monopoly II). We reverse and remand for further proceedings.

I. Prior Proceedings.

General Mills is the successor to Parker Brothers, Inc., which had produced and sold a game it called Monopoly since 1935. Parker Brothers registered “Monopoly” as a trademark in that year. In 1973 Anti-Monopoly, Inc. was established to produce and sell a game it called Anti-Monopoly. General Mills claimed that this infringed its trademark. This action was then brought by Anti-Monopoly, seeking a declaratory judgment that the registered trademark “Monopoly” was invalid, and cancelling its registration. In a counterclaim, General Mills sought declaratory and injunctive relief upholding its trademark, and the dismissal of the action. The case was tried without a jury in 1976. The court entered a judgment for General Mills. We reversed and remanded for further consideration of (i) the validity of the trademark, (ii) infringement of the trademark, if it is valid, by Anti-Monopoly, and (iii) state law claims concerning unfair competition and dilution. We also chose to defer consideration of (iv) Anti-Monopoly’s defense that General Mills had unclean hands. On remand, after hearing further evidence, the district court again entered a judgment for General Mills.

II. The Standard of Review.

We state the standard of review at the beginning of this opinion, lest we be charged with not applying it because our “acknowledgement [of it] came late in [our] opinion.” Pullman-Standard v. Swint, 1982, - U.S.-, at-, 102 S.Ct. 1781, at 1791 , 72 L.Ed.2d 66 ; id., Marshall, J., dissenting, at-, 102 S.Ct. at 1784-85 .

We must apply the standard stated in Rule 52(a), F.R.Civ.P., the case having been tried without a jury: “Findings of fact shall not be set aside unless clearly erroneous, ...” This has been interpreted to mean that the trial judge’s finding of fact cannot be set aside unless, “although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” United States v. United States Gypsum Co., 1948, 333 U.S. 364, 395 , 68 S.Ct. 525, 542 , 92 L.Ed. 746 .

The Supreme Court has recently reminded us of the importance of the rule and of our duty to abide by it. Pullman-Standard , supra,-U.S. at-, 102 S.Ct. at 1789 . In that case, the Court held that the rule applies equally to “ultimate” facts and to “subsidiary” facts. (Id.). The Court also held that, when a finding is based on an erroneous view of the law, it may be set aside, but that in such a case the appellate court cannot make a contrary finding, but must remand to the trial court for new findings, to be made in the light of the correct rule of law. (Id. at-, 102 S.Ct. at 1791-92 ). To this there is one exception: A remand is unnecessary if “the record permits only one resolution of the factual issue.” (Id., citing Kelley v. Southern Pacific Co., 1974, 419 U.S. 318, 331-332 , 95 S.Ct. 472, 479-80 , 42 L.Ed.2d 498 .) See also Inwood Laboratories v. Ives Laboratories, 1982,-U.S.-,---, 102 S.Ct. 2182, 2188-90 , 72 L.Ed.2d 606 , and the concurrence of Justice Rehnquist:

III. The Burden of Proof.

The district court ruled that Anti-Monopoly had the burden of showing genericness “by convincing evidence.” Anti-Monopoly II, 515 F.Supp. at 451-452 . The case cited for that proposition, Feathercombs, Inc. v. Solo Products Corp., 2 Cir., 1962, 306 F.2d 251 , does not announce such a rule. The only reference to “convincing evidence” is at 306 F.2d 256 , and says nothing about burden of proof. There is a presumption in favor of a registered trademark, and the burden of proof is upon one who attacks the mark as generic, but the presumption can be overcome by a showing by a preponderance of the evidence that the term was or has become generic. See Vuitton et Fils S. A. v. J. Young Enterprises, 9 Cir., 1981, 644 F.2d 769, 775-776 .

IV. Generic Terms — The Law.

Our opinion in Anti-Monopoly I binds both this court and the district court. There, we set out the law about generic terms and explained how it was to be applied to the particular facts of this case. Anti-Monopoly I, 611 F.2d at 300-306 . In this opinion, we assume that the reader will be familiar with that opinion. Here, we emphasize what we consider to be its essence. A word used as a trademark is not generic if “the primary significance of the term in the minds of the consuming public is not the product but the producer.” Id. at 302 . “[W]hen a trademark primarily denotes a product, not the product’s producer, the trademark is lost.” Id. at 301 . A registered mark is to be cancelled if it has become “the common descriptive name of an article,” 15 U.S.C. § 1064 (c), and no incontestable right can be acquired in such a mark. 15 U.S.C. § 1065 (4). We said “Even if only one producer — Parker Brothers — has ever made the MONOPOLY game, so that the public necessarily associates the product with that particular producer, the trademark is invalid unless source indication is its primary significance.” Anti-Monopoly I, 611 F.2d at 302 . “It is the source-denoting function which trademark laws protect, and nothing more.” Id. at 301 . “[0]ne competitor will not be permitted to impoverish the language of commerce by preventing his fellows from fairly describing their own goods.” Id., quoting Bada Co. v. Montgomery Ward & Co., 9 Cir., 1970, 426 F.2d 8, 11 . “[W]hen members of the consuming public use a game name to denote the game itself, and not its producer, the trademark is generic and, therefore, invalid.” Id. at 304:

V. Was the Term “MONOPOLY” Generic at the Time of Registration?

Anti-Monopoly, Inc. claims that the term “Monopoly” was generic at the time when Parker Brothers registered it. On this question, the trial judge made the following findings:

515 F.Supp. at 451-452 . The district court found also that Darrow was the inventor of the game (Id. at 451) and that the game was “created” by Darrow. Id. at 452 n.l. (The quotation marks are in the original.)

We have already held that the district court placed too heavy a burden on Anti-Monopoly, Inc. Moreover, the court’s reference to Darrow as the inventor or creator of the game is clearly erroneous. The record shows, as we stated in Anti-Monopoly I, that “The game of ‘Monopoly’ was first played from 1920 to 1982 on various college campuses by a small group of individuals, many of whom were related by blood or marriage. In late 1932 or early 1933 one of these players introduced Charles Darrow to the game, and gave him a handmade game board, rules, and associated equipment. Immediately thereafter Darrow commenced commercially producing and selling ‘Monopoly’ game equipment.” 611 F.2d at 299 .

We have re-examined the entire record on appeal. Here is what it shows. At some time between 1904 and 1934, the game of monopoly developed. Early equipment was handmade and copied from earlier handmade equipment. All the witnesses presented by Anti-Monopoly insisted that the game was known as “Monopoly” by all who played it, although in most cases the name did not appear on the board itself. The game was played in Reading, Pennsylvania, sometime between 1911 and 1917, but this date may be a little early. In the early 1920’s the game was played at Princeton University, Massachusetts Institute of Technology, Smith College, the University of Pennsylvania, and Haverford College. On occasion the rules were privately printed. The game was offered to, but rejected by, Milton-Bradley, a leading competitor of Parker Brothers. It was played in and around Reading, Pennsylvania from the early 1920’s to the early 1930’s. It may have been brought there from the University of Pennsylvania. Players in Reading made up and sold some half dozen sets of equipment at Williams College and the University of Michigan. The game next ap-. peared in Indianapolis, where some players marketed it under the name “Finance.” (Players in Reading sold some of those games too.) The game of monopoly was brought to Atlantic City, New Jersey in 1931 or thereabouts. The street names used in the game were then changed to Atlantic City street names. The game was taught to Darrow. He sold it to Parker Brothers in 1935, claiming that it was his own invention. Parker Brothers also bought the Finance game from its owners.

It is true that Darrow, in his correspondence with Parker Brothers, claimed to have invented the game and offered to sign an affidavit stating his story. However, Robert B. M. Barton, the former President of Parker Brothers, who negotiated with Darrow in 1935, testified that he did not believe Darrow’s claim. A precursor of monopoly, the Landlord’s Game, was patented by Mrs. Maggie Elizabeth Phillips of Washington, D.C. in 1904 and again in 1924. Parker Brothers purchased this game from her in 1934 or thereabouts.

In 1957, Barton, the President of Parker Brothers, in a letter to an inquirer, wrote: “So far as we know The Landlord’s Game, invented by Mrs. Elizabeth Maggie Phillips of Washington, D.C., was the basic game for both FINANCE and MONOPOLY. Mrs. Phillips patented her game and we purchased her patent. Mr. Charles Darrow later made many improvements in The Landlord’s Game and called his game MONOPOLY. He, too, secured a patent which he assigned to us. Because of the fact that we purchased all three games, it does not make very much difference to us who invented either one of the games.... ”

The evidence clearly shows that the game of monopoly was played by a small number of people before Darrow learned of it, and that these people called the game “monopoly.” It is unclear just how many people played. General Mills offered testimony that it was not widespread throughout the United States. The burden of proof was on Anti-Monopoly to show that the term was generic. We cannot presume that the evidence offered by Anti-Monopoly is the tip of the iceberg. Thus, we are faced with the following legal question: if a game is known about by a small number of people and they all call it by a particular term, may one member of the group appropriate that name by registering it as a trademark?

Ordinarily, because the trial judge imposed too heavy a burden on Anti-Monopoly, we would be required to remand for new findings made under the proper burden. However, because we have concluded that the evidence, construed most favorably to Anti-Monopoly, does not show that “Monopoly” had become generic before Parker Brothers registered it as its trademark, and because our holding in Part VI, post, is dispositive of the case, such a remand is unnecessary here.

When a small number of people use a particular thing and call it by a particular name, one which is not a common descriptive term for the thing, a person may appropriate the name and register it as a trademark. The purpose of the doctrine that generic terms cannot be made trademarks is to prevent the appropriation of a term that is already in wide use among those who are potential purchasers of the thing that the term describes. If those who might purchase the thing know it by a particular name, then to forbid the use of that name by potential producers will erect unwarranted barriers to competition. As we said in Anti-Monopoly I, “Trademarks ... are not properly used as patent substitutes to further or perpetuate product monopolies,” 611 F.2d at 300 . On the other hand, where, as here, the potential market is nationwide, and where the name is used only by a small number of scattered consumers, appropriation of the name as the trademark of one who produces for that potential market does not restrain competition to a significant degree.

We agree with the trial judge’s conclusion that “Monopoly” had not become generic before Parker Brothers registered it as a trademark.

VI. Has “Monopoly” Become Generic Since It Was Registered?

This question is discussed, and the trial court’s findings of fact appear in Monopoly II, 515 F.Supp. at 452-455 . Under the heading “FINDINGS OF FACT,” the following appears:

Id. at 455 . We consider finding 2 to be one of ultimate fact, and subject to the “clearly erroneous” standard of Rule 52(a). See Pullman-Standard, supra,-U.S. at-, 102 S.Ct. at 1789.

The district court also said “ ‘Primary significance’ logically implies a hierarchical priority over a competing alternative.” 515 F.Supp. at 454 . Dictionary definitions are in accord. Funk & Wagnalls’ New Standard Dictionary gives “primary 1. First in ... thought or intention, 2. First in degree, rank or importance, most fundamental, chief....” Webster’s New International Dictionary (2d Ed.) gives “1. First in ... intention; 2. First in ... importance; chief, principal.... ” We are not sure what the district court meant by a “competing alternative.” To us, this carries some suggestion of “either, or.” Yet it is nearly always the case, as the district court recognized, that a trademark will identify both the product and its producer. Anti-Monopoly II, 515 F.Supp. at 454 . Indeed, its value lies in its identification of the product with its producer.

In its opinion, the district court supports its finding 2 as follows:

Id. at 454-455 (emphasis in the original).

In considering whether these findings, and finding 2, are clearly erroneous, we have in mind an obvious proposition. The word “Monopoly,” while not in its ordinary meaning descriptive of the game “Monopoly,” is an ordinary English word, and it does describe the objective of the game. This was recognized in the rules of the game published by Parker Brothers in 1935. They begin with:

A Monopolist has a monopoly. By choosing the word as a trademark, Parker Brothers subjected itself to a considerable risk that the word would become so identified with the game as to be “generic.”

In Anti-Monopoly II the district court also said this: “Unless the Ninth Circuit standard is meant to foreclose the possibility of trademark protection for any producer of a unique game whose corporate name does not appear in the title of the game (e.g., ‘SCRABBLE,’ ‘TOWER OF BABBLE’), then its test cannot be used here to thwart MONOPOLY’S trademark rights.” 515 F.Supp. at 455 . Nothing in our opinion in Anti-Monopoly I even hints at the relevance of whether or not the corporate name of the producer of a game appears in the title of the game, and our opinion does not foreclose the possibility of trademark protection of the name of a game that does not embody the corporate name of its producer. But our opinion does squarely hold as follows: “Even if only one producer — Parker Brothers — has ever made the MONOPOLY game, so that the public necessarily associates the product with that particular producer, the trademark is invalid unless source identification is its primary significance.” 611 F.2d at 302 .

The district court obviously felt that our opinion in Anti-Monopoly I gave Anti-Monopoly an easier task in trying to show that “Monopoly” has become generic than the district court would give. Nevertheless, both we and the district court are bound by our decision in Anti-Monopoly I.

We now consider whether finding 2 of the district court is clearly erroneous. We conclude that it is.

As we have seen, the district court relied in part upon the fact that General Mills and its predecessor have spent time, energy, and money in promoting and policing use of the term “Monopoly.” That fact, however, is not of itself sufficient to create legally protectable rights. HMH Publishing Co. v. Brincat, 9 Cir., 1974, 504 F.2d 713, 719 . It is not, of itself, enough that over 55% of the public has come to associate the product, and as a consequence the name by which the product is generally known, with Parker Brothers. Anti-Monopoly I, 611 F.2d at 302 . See also Kellogg Co. v. National Biscuit Co., 1938, 305 U.S. 111, 118 , 59 S.Ct. 109, 113 , 83 L.Ed. 73 (the “Shredded Wheat” case). Even if one third of the members of the public who purchased the game did so because they liked Parker Brothers’ products, that fact does not show that “Monopoly” is primarily source indicating. The very survey on which the district court placed emphasis by italicizing its result shows that two thirds of the members of the public who purchased the game wanted “Monopoly” and did not care who made it.

The real question is what did Parker Brothers and General Mills get for their money and efforts? To us, the evidence overwhelmingly shows that they very successfully promoted the game of Monopoly, but that in doing it they so successfully promoted “Monopoly” as “the name of the game,” that it became generic in the sense in which we use that term in trademark law. We recognize that “there is evidence to support” the trial court’s findings, United States Gypsum Co., supra, 333 U.S. at 395 , 68 S.Ct. at 542 , but “on the entire evidence [we are] left with the definite and firm conviction that a mistake has been committed.” Id.

The principal evidence in the case was in the form of consumer surveys, and to these we now turn.

A. The Brand-name Survey.

General Mills conducted a survey based upon a survey approved by a district court in the “Teflon” case, E. I. Du Pont de Nemours & Co. v. Yoshida International, Inc., E.D.N.Y., 1975, 393 F.Supp. 502 . In the survey conducted by General Mills, people were asked whether “Monopoly” is a “brand-name,” and were told: “By brand name, I mean a name like Chevrolet, which is made by one company; by common name, I mean ‘automobile,’ which is made by a number of different companies.” (Emphasis in the original.) The results of this survey had no relevance to the question in this case. Under the survey definition, “Monopoly” would have to be a “brand name” because it is made by only one company. This tells us nothing at all about the primary meaning of “Monopoly” in the minds of consumers.

It is true that the witness through whom the survey was introduced testified on direct examination that as a result of it his opinion was that “Monopoly” primarily denotes source or producer. However, on cross-examination and redirect examination it became clear that this witness had done no more than reduplicate the “Teflon” survey (with appropriate substitutions and slight additions) and had no opinion on the relevance of this survey to any issue in the present case. The brand-name survey is not even some evidence to support finding 2; it is no evidence to support it.

B. The “Thermos” Survey.

Anti-Monopoly’s first survey was based upon that used in the “Thermos” case, King-Seeley Thermos Co. v. Aladdin Industries, Inc., D.Conn., 1962, 207 F.Supp. 9, 20-21 , aff’d, 2 Cir., 1963, 321 F.2d 577 . In Anti-Monopoly’s survey people were asked the question: “Are you familiar with business board games of the kind in which players buy, sell, mortgage and trade city streets, utilities and railroads, build houses, collect rents and win by bankrupting all other players, or not”? About 53% said they were. Those people were then asked: “If you were going to buy this kind of game, what would you ask for, that is, what would you tell the sales clerk you wanted”? About 80% said: “Monopoly.”

The witness through whom this survey was introduced testified that Anti-Monopoly gave his firm the questions used in the “Thermos” survey and asked it to conduct a similar one. Anti-Monopoly provided the wording of the questions in the present survey as well. The research firm was responsible for deciding how to reach a sample that would adequately represent the population of the United States. The witness gave no testimony as to the relevance of the results of the survey to the issues in the case.

In one of its briefs, General Mills points out that the survey used in the “Thermos” case was described as “generally corroborative of the court’s conclusions drawn from other evidence,” and that the district court which decided the “Teflon” case found a “Thermos”-like survey defective because “the design of the questions more often than not [focused] on supplying the inquirer a ‘name,’ without regard to whether the principal significance of the name supplied was ‘its indication of the nature or class of an article, rather than an indication of its origin,’ King-Seeley Thermos Co., supra, 321 F.2d at 580 .” E. I. Du Pont de Nemours & Co., supra, 393 F.Supp. at 527 . Be that as it may, we think that the results of this survey are compelling evidence of a proposition that is also dictated by common sense: an overwhelming proportion of those who are familiar with the game would ask for it by the name “Monopoly.”

C. The Motivation Survey.

After the remand to the district court, Anti-Monopoly commissioned a further survey. This survey was based upon the following language from our opinion in Anti-Monopoly I:

611 F.2d at 305-306 . The wording of the questions was provided by Dr. Anspach, Anti-Monopoly’s president, and by the expert who testified at trial. The expert had studied our first opinion. The survey was designed to ascertain the use of the term “Monopoly” by those who had purchased the game in the past or intended to do so in the near future. It was conducted by telephone. The results were as follows: 92% were aware of “Monopoly,” the business board game produced by Parker Brothers. Of that 92%, 62% either had “purchased ‘Monopoly’ within the last couple of years” or intended to purchase it in the near future. Those people were asked why they had bought or would buy monopoly. The answers exhibited the following pattern: 82% mentioned some aspect of the playing of the game (e.g., that they played it as a kid, it was a family game, it was enjoyable, it was fun to play, it was interesting), 14% mentioned some educational aspect of the game, 7% mentioned the equipment (e.g., saying it was durable) or said they were replacing a set, 1% spoke of price, 34% gave other reasons neutral to the issues in this case (e.g., it was for a gift, the game was a classic, people like the game). The percentages total more than 100 because respondents often gave more than one reason.

The people who said that they had purchased the game within the last couple of years or would purchase it in the near future were then given a choice of two statements and were asked which best expressed their reasons. Sixty-five percent chose: “I want a ‘Monopoly’ game primarily because I am interested in playing ‘Monopoly,’ I don’t much care who makes it.” Thirty-two percent chose: “I would like Parker Brothers’ ‘Monopoly’ game primarily because I like Parker Brothers’ products.”

A very similar “intercept survey” was conducted by face to face interviews. The results were very close to those of the tele phone survey, but the expert did not claim that the intercept study was validly project-able.

The district court indicated its reasons for rejecting this survey. Insofar as these are findings of fact they must be accorded the deference required by Rule 52(a), and, giving them that deference, we hold them to be clearly erroneous. The district court’s major objection to the survey was that it sought an explanation of an actual purchaser’s motivation in purchasing the game rather than the primary significance of the word. Anti-Monopoly, Inc., 515 F.Supp. at 453 . This objection cannot stand. In our earlier opinion we made it clear that what was relevant was the sense in which a purchaser used the word “Monopoly” when asking for the game by that name. The survey was a reasonable effort to find that out and was modelled closely on what we said in our opinion.

The district court thought that the survey was invalidated by the fact that in the first question people were asked if they were “aware of ‘Monopoly,’ the business board game produced by Parker Brothers” (emphasis supplied). It supposed that the presence of the emphasized words somehow inhibited those who might otherwise have responded to later questions that they bought the game because it was produced by Parker Brothers. No evidence or expert opinion was given to support this view and it has no inherent plausibility.

In a footnote the district court said of this survey that “other methodological deficiencies abound.” Id., 515 F.Supp. at 453 n.4. One suggested deficiency is that Professor Anspach suggested the language that was used. This is taken to be evidence of “inherent bias.” General Mills argues to us that little weight should be given to this survey because it was devised by Dr. An-spach and the survey firm without the mediation of a trademark attorney. We find no merit in these objections.

The district court found that the study was “overwhelmingly prone to errors of subjective grading.” Id. No doubt it was referring to the process by which responses were categorized as, for example, education, enjoyable, “played it as a kid,” or equipment. This process of categorization was not purely mechanical, and did involve some use of human judgment. However, we are not prepared to dismiss every process that includes the operation of human judgment as “overwhelmingly prone to errors of subjective grading.” Nor do we find any special reasons to suspect the exercise of judgment in the case of this survey. The categories that were listed strike us as reasonable ones.

Neither the district court nor General Mills claims that there were in fact errors of judgment, but only that there might have been. The raw responses to the survey were at one point offered in evidence by Anti-Monopoly, but the offer was withdrawn after General Mills objected, citing F.R. Evidence 705, 1005 and 1006, and the district judge said: “if [counsel for General Mills had] asked for them, he could have received them. If he received them, he could have turned them over to his expert to check them out and see if they give a reliable or non-reliable basis for the opinion. But since he didn’t ask for them, I don’t think they should go into evidence.” Under these circumstances, General Mills cannot now argue that the raw responses were not in fact correctly categorized.

Finally, in the same footnote, the district court suggested that the result that 82% of monopoly purchasers buy for “product related” reasons cannot be reconciled with the other result that 32% of actual or potential buyers chose the statement “I would buy Parker Brothers’ ‘monopoly’ game primarily because I like Parker Brothers’ products.” This is a misconception of the survey results. The comparable figure to the 32% is the 65% who chose the statement “I want a ‘monopoly’ game primarily because I am interested in playing ‘monopoly,’ I don’t much care who makes it.” The 82% who gave “product related” answers no doubt had both product related and source related reasons for buying, and, with some, enough to reduce 82% to 65%, the source related reason was stronger when the person had to choose. But it is still true that 65% chose product, rather than source.

We conclude that the findings regarding the survey are clearly erroneous, and that it does support the conclusion that the primary significance of “Monopoly” is product rather than source.

D. The Tide Survey.

General Mills introduced a survey that was intended as a reductio ad absurdum of the motivation survey. It showed that when asked to supply a reason for buying Tide about 60% of those who might buy it now or in the future said that they would buy Tide because it does a good job. However, when asked “Would you buy Tide primarily because you like Procter and Gamble’s products, or primarily because you like Tide detergent?” about 68% indicated the latter reason. There were various respects in which this survey was different from the motivation survey used by Anti-Monopoly, but we shall not suddenly attach great importance to technical considerations. We suspect that these results tend to show that the general public regards “Tide” as the name of a particular detergent, having particular qualities, rather than as one producer’s brand name for the same detergent which is available from a variety of sources. We do not know whether the general public thinks this, or if it does, is correct in thinking this, or whether Procter and Gamble intend them to think it. If the general public does think this, and if the test formulated in Anti-Monopoly I could be mechanically extended to the very different subject of detergents, then Procter and Gamble might have cause for alarm. The issue is not before us today. The motivation survey conducted by Anti-Monopoly, Inc. was in accordance with the views we expressed in Anti-Monopoly I. The results in the Tide Survey are of no relevance to this case.

E. Conclusion.

We hold that Finding 2 is clearly erroneous because, although there is some evidence to support it, our examination of the evidence leaves us with the definite and firm conviction that a mistake has been committed. We hold that, as applied to a board game, the word “Monopoly” has become “generic,” and the registration of it as a trademark is no longer valid.

VII. Other Issues.

The district court must determine whether Anti-Monopoly is taking reasonable care to inform the public of the source of its product, and if it finds that this is not so may enjoin the sale of anti-monopoly save upon appropriate conditions. Anti-Monopoly I, 611 F.2d at 307 .

We remand the case. The district court shall enter judgment for Anti-Monopoly, Inc. on the question of trademark validity and take whatever actions are necessary and consistent with this opinion.