W.T. Rogers Company, Inc. v. Wendell R. Keene and Keene Manufacturing, Inc.

778 F.2d 334, 228 U.S.P.Q. (BNA) 145, 1985 U.S. App. LEXIS 25337, 54 U.S.L.W. 2344

• Court: Court of Appeals for the Seventh Circuit
• Decided: November 26, 1985
• Docket: 85-1208
• Status: Published

Opinion

POSNER, Circuit Judge.

A brand’s shape, color, pattern, or other design characteristic cannot obtain the protection of trademark law if it is “nonfunctional”; and we are required in this case to decide what the standard of functionality shall be in this circuit.

The plaintiff, Rogers, and the corporate defendant, Keene, are competing manufacturers of office supplies; Wendell Keene, the individual defendant, is the founder and president of the corporation. Both companies are located in Wisconsin. In 1969 Rogers began to manufacture a molded plastic stacking office tray for letters and other documents. As shown in the photograph at the end of this opinion, the sides of the tray are hexagonal, and have little holes on the top and “feet” on the bottom so that the tray can be clamped together with other trays to form a stack. The Rogers tray was for long the only tray with hexagonal end panels; competing trays had rectangular panels. Whether because the hexagonal design was pleasing to customers or for other reasons, the Rogers tray was a big success and today Rogers sells about one million of the trays each year.

Mr. Keene was an officer, and later a sales representative, of Rogers before he founded his own company in 1983. When he went to a plastics firm to get a mold made for the plastic stacking office tray that he wanted to manufacture, he gave the molder a Rogers tray, and asked him to make a mold for a similar tray. In fact the tray that the molder produced and that the Keene corporation sells as its “Model 1001” is virtually identical to the Rogers tray. Before Keene set up in competition with Rogers, Rogers hadn’t worried much about affixing a trade name to its tray, but afterward it stamped “Stak-Ette” on the tray.

Rogers never tried to get a design patent for the hexagonal end panels that are the most distinctive feature of its tray; nor did it ever try to register the feature as a trademark. It claims nevertheless to have a common law trademark in the feature, which it contends identifies plastic stacking office trays made by Rogers. It brought this suit under section 43(a) of the Lanham Act, 15 U.S.C. § 1125(a), which creates a federal remedy for making “a false designation of origin, or any false description or representation, including words or other symbols tending falsely to describe or represent the same____” Since the purpose of a trademark, whether federally registered or unregistered, is to designate the origin of goods, the infringement of such a trademark is actionable under section 43(a), provided the other requirements of the section are met. See, e.g., Bose Corp. v. Linear Design Labs, Inc., 467 F.2d 304, 309-10 (2d Cir.1972); 1 Gilson, Trademark Protection and Practice § 2.13[4] (1984); 2 McCarthy, Trademarks and Unfair Competition § 27:3 (2d ed. 1984). And provided that a defense of functionality is recognized, there is no conflict with federal patent law, save possibly with 35 U.S.C. § 171, which allows a 14-year patent to be granted for a nonfunctional ornamental design — a design patent. But the courts that have considered the issue have concluded, rightly in our view, that this section does not prevent the enforcement of a common law trademark in a design feature. See discussion in 1 Chi-sum, Patents § 1.04[6] (1985). The trademark owner has an indefinite term of protection, it is true, but in an infringement suit must also prove secondary meaning and likelihood of confusion, which the owm er of a design patent need not do; there is therefore no necessary inconsistency between the two modes of protection. Cf. Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470, 489-91, 94 S.Ct. 1879, 18889-91, 40 L.Ed.2d 315 (1974); Aronson v. Quick Point Pencil Co., 440 U.S. 257, 99 S.Ct. 1096, 59 L.Ed.2d 296 (1979).

We have said, and the plaintiff does not deny, that functionality is a defense to a suit under section 43(a); but perhaps a few words are in order about the source of the defense. Functionality is a traditional defense in the common law of unfair competition to a suit for trademark infringement. One view of section 43(a) is that the section merely provides a féderal remedy for unfair competition (occurring in interstate commerce), and on that view we would look to the common law of Wisconsin for the contours of the functionality defense. Another view is that section 43(a) creates a federal substantive law of unfair competition in interstate commerce, or more precisely authorizes the federal courts to create such a law; in other words, that it is a grant of common law powers to these courts. On this view a suit such as this is a suit to enforce a federal common law trademark, and the court is not bound to follow the common law of a particular state. Perhaps therefore it would not be bound to recognize a defense of functionality; perhaps, indeed, the rule should be that functional features can be protected where necessary to prevent confusion unless competition is unduly hindered, see Note, The Problem of Functional Features: Trade Dress Infringement Under Section 43(a) of the Lanham Act, 82 Colum.L.Rev. 77, 81-93 (1982). This leaves out of account, however, the importance of recognizing a defense of functionality in order to head off a collision between section 43(a) and patent law. There is also the considerable practical objection, reminiscent of the objections to the general federal common law of the Swift v. Tyson era, that confusion, complexity, and forum-shopping would be the result of allowing firms to enforce simultaneously, in different courts, state and federal common law trademarks.

We shall not have to decide whether the common law trademark that section 43(a) protects is state or federal, and whether, if the former, still the defense of functionality is federal; the parties do not suggest that it would make any difference to the outcome of this case how the questions are answered. We mention them merely to make clear that they are questions that this opinion does not resolve, implicitly or otherwise.

The case was tried to a jury, which was given a special verdict to fill in that required it to answer (so far as relevant to this appeal) the following sequence of questions. The first was whether the hexagonal end panels on the Rogers tray are nonfunctional. If the jury answered “no,” it was not to answer the following questions. They were whether the hexagonal end panel, if nonfunctional, either identified the Rogers tray or (what amounts to the same thing, but was confusingly stated as an alternative) had a “secondary meaning” (that is, whether it signified to consumers the Rogers tray), and if so whether Keene’s duplicating the design for its own tray was likely to confuse consumers. The jury answered the first question “no” and therefore stopped; the district judge then entered judgment for Keene.

As should be apparent from the structure of the special verdict, the concept of functionality is intended to screen out from the protection of trademark law certain design features even if they have become so far identified with the manufacturer of a particular brand that consumers may be confused about the origin of the good if another producer is allowed to adopt the feature. The reason for this screen or filter becomes apparent once the purpose of trademark protection is understood. (For useful background see Note, Promotional Goods and the Functionality Doctrine: An Economic Model of Trademarks, 63 Tex.L.Rev. 639, 656-62 (1984).) The purpose is to reduce the cost of information to consumers by making it easy for them to identify the products or producers with which they have had either good experiences, so that they want to keep buying the product (or buying from the producer), or bad experiences, so that they want to avoid the product or the producer in the future. Scandia Down Corp. v. Euroquilt, Inc., 772 F.2d 1423, 1429-30 (7th Cir.1985). This purpose is achieved by letting a producer pick an identifying name or symbol for his brand, and forbidding competing producers to use the same or a confusingly similar name or symbol on their brands. Since the supply of distinctive names and symbols usable for brand identification is very large, indeed for all practical purposes infinite, competition is not impaired by giving each manufacturer a perpetual “monopoly” of his identifying mark; such marks are not a scarce input into the production of goods. See In re Mogen David Wine Corp., 328 F.2d 925, 933 (C.C.P.A.1964) (Rich, J., concurring). But if instead of picking some distinctive mark for his brand, a manufacturer tries to appropriate the word that consumers use to describe the entire product (comprising his brand plus all competing brands) — for example, tries, if he is a manufacturer of automobiles, to use “automobile” as his trademark, rather than “Chevrolet” — then he is trying to monopolize a scarce input, for there usually are only one or two words in common usage to describe a given product (such as “car” and “automobile”). In such a case trademark protection is denied. Scandia Down Corp. v. Euroquilt, Inc., supra, 772 F.2d at 1430. In the language of trademark lawyers, “generic” names may not be trademarked.

The same principle applies if the trademark is part of the design of the product (its shape, pattern, etc.) rather than a word or logo affixed to the product. If the feature is ornamental, fanciful, decorative, like the patterns on a piece of china or of silverware, then the manufacturer can use it as his name, his symbol, his identifying mark. Ornamental, fanciful shapes and patterns are not in short supply, so appropriating one of them to serve as an identifying mark does not take away from any competitor something that he needs in order to make a competing brand. But if the feature is not ornamental or fanciful or whimsical or arbitrary, but is somehow intrinsic to the entire product consisting of this manufacturer’s brand and his rivals’ brands, trademark production will be denied. The name of this principle is “functionality,” on which see, e.g., In re Morton-Norwich Products, Inc., 671 F.2d 1332, 1338-41 (C.C.P.A.1982). Thus the first company to make an airplane cannot use the characteristic shape of an airplane as its trademark, thereby condemning its rivals to build airplanes that won’t fly. Cf. Pope Automatic Merchandising Co. v. McCrum-Howell Co., 191 Fed. 979, 981-82 (7th Cir.1911). A firm that makes footballs could not use as its trademark the characteristic oval shape of the football, thereby forcing its rivals to find another shape for their footballs; since they wouldn’t be able to sell any round or oblong or hexagonal footballs, that firm would have, not an identifying mark, but a product monopoly, and a product monopoly not for a term of years as under the patent laws but forever.

The football’s oval shape is “functional” in the following practical sense: it would be found in all or most brands of the product even if no producer had any desire to have his brand mistaken for that of another producer. A feature functional in this sense — a feature that different brands share rather than a feature designed to differentiate the brands — is unlike those dispensable features of the particular brand that, like an arbitrary identifying name, rivals do not need in order to compete effectively. So if an automobile manufacturer places at the front end of its hood a statue of Mercury, it can if it wants make this its trademark (or one of its trademarks), because its competitors do not need a statue of Mercury on the hoods of their cars in order to be able to compete. To put this differently, a functional feature is one which competitors would have to spend money not to copy but to design around, as they would have to do if they wanted to come up with a nonoval substitute for a football. It is something costly to do without (like the hood itself), rather than costly to have (like the statue of Mercury). See, e.g., Kellogg Co. v. National Biscuit Co., 305 U.S. 111, 122, 59 S.Ct. 109, 115, 83 L.Ed. 73 (1938); Inwood Laboratories, Inc. v. Ives Laboratories, Inc., 456 U.S. 844, 850 n. 10, 102 S.Ct.2182, 2187, 72 L.Ed.2d 606 (1982) (dictum). “If the feature asserted to give a product distinctiveness is the best, or at least one, of a few superior designs for its de facto purpose, it follows that competition is hindered,” In re Bose Corp., 772 F.2d 866, 872 (Fed.Cir. 1985), and trademark protection will be denied.

The difficult cases, and this is one, are cases in which the feature sought to be trademarked can be said to be functional only if giving aesthetic pleasure is a function. (For a very full discussion of the cases, see Duft, “Aesthetic” Functionality, 73 Trademark Rptr. 151 (1983).) Suppose we say that one function of an automobile is to be visually impressive, opulent, magnificent — in a word, beautiful; and that the little statue of Mercury is not just an identifying symbol but one of the details, one of the adornments, that make this manufacturer’s automobile a thing of beauty; that it is like the fluting in a Corinthian column, which is unrelated to the support as distinct from aesthetic function of a column (today many columns have no support function, but are purely aesthetic). If this view were taken, functionality would swallow up much, perhaps all, of trademark law. Rare is the manufacturer who will not try to choose a pleasing name, symbol, or design feature as his trademark. In an age when fashion-conscious consumers wear T-shirts emblazoned with the trademarks of consumer products and owners of Volkswagens buy conversion kits to enable them to put a Rolls Royce grille on their car, it is apparent that trade names, symbols, arid design features often serve a dual purpose, one part of which is functional in the sense of making the product more attractive, and is distinct from identifying the manufacturer or his brand to the consumer.

Thus we find cases such as Wiley v. American Greetings Corp., 762 F.2d 139 (1st Cir.1985), where the plaintiff manufactured a teddy bear with a bright-red heart-shaped blazon over the teddy bear’s heart, and the First Circuit held that the heart could not be appropriated as a trademark. The ground was that the blazon was not sufficiently distinctive to be a trademark but another way of analyzing it would be that it went to enhance the function of the teddy bear, which is to please children; so viewed, Wiley would be a case resting on the razor’s edge that separates the functional and the nonfunctional. Another hard case is Industria Arredamenti Fratelli Saporiti v. Charles Craig, Ltd., 725 F.2d 18 (2d Cir.1984), where the distinctive modular design of a sofa was held to be functional because “this overall design [which the defendant had copied] makes jthe sofas attractive to buyers despite enor'mously expensive price tags,” and “is not just a label; it is a principal characteristic of the sofa.” Id. at 20 (footnote omitted).

It is doubtful that any simple rule could be devised to decide these cases. On the one hand it would be unreasonable to deny trademark protection to a manufacturer who had the good fortune to have created a trade name, symbol, or design that became valued by the consuming public for its intrinsic pleasingness as well as for the information it conveyed about who had made the product, unless the feature in question had become generic, and therefore costly to engineer around; and we reject the contrary intimations in some cases from the Ninth Circuit, see, e.g., International Order of Job’s Daughters v. Lindeburg & Co., 633 F.2d 912, 918-20 (9th Cir. 1980), as have most other courts. See, e.g., In re DC Comics, Inc., 689 F.2d 1042, 1049-50 (C.C.P.A.1982) (Rich, J., concurring), and cases cited there. But it would also be unreasonable to let a manufacturer use trademark law to prevent competitors from making pleasing substitutes for his own brand; yet that would be the effect of allowing him to appropriate the most pleasing way of configuring the product. For a good discussion see Keene Corp. v. Paraflex Industries, Inc., 653 F.2d 822, 825-27 (3d Cir.1981).

Our legal system often gives the very difficult cases to juries to decide. If it seems perverse to give laymen the hard cases (assuming there is a right to jury trial, and a party who wants to exercise the right) and to give judges, who are legal professionals, only the clear cases to decide (which they do by granting summary judgments, or motions to dismiss, or motions for directed verdict or for judgment notwithstanding the verdict), the other side of the coin is that it just does not matter that much how the very close case is decided. Either way the result would be reasonable. That is why a case is called close; the only important thing is to prevent unreasonable results by putting checks on the jury’s discretion. But of course the parties to the very close case care passionately how it is decided; and Rogers argues that it should have a new trial because the jury that decided that the hexagonal end panels on the Rogers tray are functional was misinstructed on the issue.

Here is the instruction that the judge gave on functionality:

The law provides that design features that are functional are in the public domain and may be copied or simulated by others. Therefore, the burden in this first question is on the plaintiff to prove that the design of the end panel of its letter tray is nonfunctional.

The term nonfunctional means that the primary purpose served by a design feature is to identify the manufacturer or distinguish the product from that of competitors as opposed to any other purpose. That is, in order to be nonfunctional, the end panel design of the plaintiff’s plastic letter tray must primarily serve to identify it as the product of a particular company. Even if a design feature serves some other purpose, the feature is still nonfunctional if the other purpose is merely incidental.

An important component of this question is whether the appropriation of a design feature by the first user will hinder competition. For instance, a particular feature that is one of only a limited number of options cannot be considered nonfunctional because to do so would limit the number of entrants into the market.

A feature can be considered functional when it serves to provide a reason for purchase which is unrelated to the fact that the source of the product is a particular manufacturer.

Accordingly, you must decide the question whether the end panel design of the plaintiff’s plastic letter tray is nonfunctional by weighing its impact as a mark of identification against any useful purposes it might also serve. You may answer the first question yes only if the design serves primarily as a mark of identification.

The instruction contains three errors. The first is to (repeatedly) define nonfunctional as serving primarily to identify the manufacturer. Understood literally, this would mean that if a particular design feature had two equally important purposes, one to please consumers and the other to identify the manufacturer, it would be functional and could not be trademarked. But a trademark, especially when it is part of the product, rather than being just the brand name, is bound to be selected in part to be pleasing; so this definition of functionality could rule out trademark protection for design features. The second error is the statement, “A feature can be considered functional when it serves to provide a reason for purchase which is unrelated to the fact that the source of the product is a particular manufacturer.” A reason — not the most important or even equally important reason; hence a pleasing trade name, symbol, or design feature cannot be trademarked. This is wrong. See LeSportsac, Inc. v. K Mart Corp., 754 F.2d 71, 77-78 (2d Cir.1985). The third error is the reference to “limiting] the number of entrants into the market.” If a market contains many competitors, reducing that number by one, or preventing one new competitor from entering, is unlikely to have any significant anticompetitive effect; so the quoted phrase does not correctly frame the competitive inquiry that the jury is required to make.

Of course the fact that a jury instruction contains errors does not by it self dictate reversal, perhaps especially in cases where the jury’s verdict is for the defendant; he who lives by the jury shall perish by it (but here the defendants requested a jury trial first). Instructions must be read as a whole with due regard for the artificiality of assuming that isolated passages in a lengthy set of instructions are apt to spell the difference between victory and defeat. But the instruction we have quoted is the only instruction the judge gave on the issue of functionality; and that issue was the only issue (relevant to this appeal) that the jury decided. Moreover, the errors we have noted, particularly the first, are not detachable from other and correct passages in the instruction, but are the heart of the instruction. This is not to say that the case would in fact have been decided differently if the instruction had been correct. At argument we asked Rogers’ counsel why he thought Rogers had lost before the jury and he answered that he thought it was because Rogers is a much bigger firm than Keene. But if we treated this statement as a confession that the errors in the instruction on functionality were harmless, we would come close to saying that the judge would not have committed reversible error if he had sent the case to the jury with no instructions at all.

It would be different if the case were so one-sided that one could predict with a high degree of confidence that the facts, and not the instructions or extraneous considerations such as the relative sizes of the parties, had determined the outcome; and Keene argues with great vigor that this was indeed the case. It points out that the evidence shows that until it began making a tray that resembled (in fact, duplicated) the Rogers tray, Rogers had made little effort to associate the hexagonal end panel 'with the name “Rogers,” and had even allowed some of its buyers to affix their own trade names to the trays before reselling them to consumers. An additional point is that a plastic stacking tray is a cheap and simple mass-produced good which is not ingested (like food) and does not have to be used to be evaluated (like a fountain pen); all its relevant qualities are immediately accessible to observation. The users of such a good may not care about the producer’s identity or even about whether they are getting the “same” tray they have bought previously. But these points are not even related to the issue whether the hexagonal end panel is functional. They amount to a prediction that if the jury had answered question number one in the special verdict “yes” it would have gone on and answered other questions “no” and Keene would have won anyway. If the evidence is so clear we wonder why Keene did not move for a directed verdict, but in any event the evidence is not so clear. Rogers may not have bothered to emphasize its name in conjunction with the hexagonal end panels merely because the buying public in fact, from the first, recognized trays so paneled as Rogers trays. There is evidence that some consumers wanted the Rogers tray — a particular brand — and there is some evidence of confusion.

The issue of functionality is distinct. It is whether the hexagonal end panel is a feature common to the entire product, that is, common to the molded plastic office stacking tray, or whether it is the kind of dispensable feature that Rogers is entitled to appropriate under trademark law. On this there is much evidence to support Rogers’ position. Before Rogers began selling its tray, molded plastic office stacking trays came with rectangular end panels rather than hexagonal ones. So evidently the hexagonal shape, like the rhomboidal shape of the candy bar involved in Application of World’s Finest Chocolate, Inc., 474 F.2d 1012 (C.C.P.A.1973), was not something intrinsic to the product, as the oval shape of the football is. Of course, products change, and if the hexagonal end panel made the tray substantially more useful, or substantially cheaper to produce, competing manufacturers would be entitled to copy it unless forbidden by patent law (which they would not be, since as we have said Rogers never sought patent protection). But there is no suggestion of that here. The hexagonal shape of the end panel does nothing to enhance the tray’s utility in holding papers; the shape is as irrelevant to that function as the fluting in a column is irrelevant to the column’s function of holding up the roof. And there is no suggestion that it makes a tray cheaper to produce; if anything, it probably makes it more expensive to produce, though probably trivially so.

It is only when the term functional is expanded to embrace the aesthetic that it becomes possible to argue that the hexagonal shape is functional. A molded plastic stacking office tray is an inherently humble product, and the opportunities for adornment may be limited. But office workers are not indifferent to their surroundings, and if the hexagonal end panel imparts a degree of elegance that makes the tray worth more, the hexagonal design performs a genuine function. That does not mean that the design cannot be trademarked, for as we said earlier an identifying mark does not get thrust into the public domain just because the buying public finds it pleasing. But suppose the only thing the hexagonal end panel does is enhance the attractiveness of the tray; suppose it has no more to do with marking the tray as a product of Rogers than the characteristic silhouette of a station wagon marks it as a product of General Motors. Rogers would deserve a reward for hitting on a pleasing design which competitors were quick to copy but it would have to seek that reward under the design-patent law rather than under trademark law. Beauty is function.

But the fact that a design feature is attractive does not, to repeat, preclude its being trademarked. If effective competition is possible without copying that feature, then, by analogy to the distinction between arbitrary and generic brand names, it is not a functional feature. That may well be the case here, since Rogers is outsold by Eldon, the maker of a rectangular tray — though of course other factors may explain Eldon’s success; Rogers’ may really be the superior design. Even if competing manufacturers of molded plastic stacking office trays could not use any form of hexagonal end panel without infringing Rogers’ trademark, that would not confine them to the rectangular end panel, which although still the commonest shape is we may assume too drab to be a good substitute for a fancier shape. For even with the hexagon appropriated an infinity of geometrical patterns would remain open to competitors. Keene could have chosen for his end panel an oval, a pentagon, a trapezoid, a parallelogram, an octagon, a rectangle covered with arabesques, or machicolated, or saw-toothed. The hexagon is not the only feasible shape for the side of a tray. Moreover, we do not understand Rogers to be claiming the hexagon. The hexagon is not a uniform shape, like a circle or a square or an equilateral triangle; it is any six-sided figure. Rogers has chosen a particular hexagon as the shape for its end panels, and a hexagon moreover with a hole in it. The ensemble may well be distinct from a number of other hexagonally shaped end panels, and if so the options of competing manufacturers may be as a practical matter unlimited.

The strongest case for Keene, outside of the Ninth Circuit cases that we and most courts believe give too little protection to pleasing trademarks, is another Keene case (not the same company, though), Keene Corp. v. Paraflex Industries, Inc., supra. There a finding that the design of the plaintiff’s “Wall Cube,” an outdoor light fixture consisting of two oblong right triangular solids, one a metal housing and the other a slightly smaller glass lens, was functional was upheld in part on the basis of the needs of architectual compatibility, which limited the variety of shapes in which such fixtures could be made. Keene argues that the décor compatibility of stacked trays requires that any manufacturer of such trays be allowed to use the same shaped end panel as any other. The premise of the argument is that a stack of trays with differently shaped ends would be ugly, so that even if each shape were perfectly arbitrary in relation to any use made of an office tray, and therefore nonfunctional, whatever shape the first manu facturer adopted would establish a norm from which other manufacturers could not depart with any hope of competitive success.

There is only superficial appeal to this point. For one thing Rogers was not the first maker of stacking trays. For another, they are cheap items and if someone came along with a more elegant design many office managers would be willing to replace an entire stack on some, perhaps all, of the desks in the office. For a third, as an office’s existing stack wears out or as new companies spring up or new offices are built a new manufacturer with a differently shaped end panel will find a ready market. Rogers was able to enter the market successfully with a tray shaped differently from the then dominant rectangular shape; why should other producers have greater difficulty?

The biggest objection to the point about décor compatibility, however, is that it is an open Sesame to trademark infringement. Suppose Mr. Keene owned a complete set of Meissen china, and one of the plates broke. He might care more about replacing it with a plate that looked exactly like the plate that had broken than about who made the plate; but it would not follow that someone could make exact duplicates of Meissen china for sale to people who care more about aesthetic compatibility than about source. A design feature to be aesthetically functional must be pleasing in itself; it is not enough that a person who owns two items with that feature wants a matched pair. Otherwise we might be forced to admit that General Motors can duplicate the Rolls Royce, because a person who had one Rolls Royce might think a second Rolls would look good next to it in his garage.

The most difficult issue in the case is whether Rogers is entitled to a new trial because of the error in the instruction, given that its own proposed instruction on functionality was also erroneous. The proposed instruction was:

The plaintiff cannot protect the symbol which it asserts is a trademark if the end panel design of its letter tray is primarily functional. You will find the design functional if you find that it is dictated by utilitarian characteristics or by the function that the relevant product is intended to serve. To be functional, the design or feature must be found superior or optimal in terms of engineering, economy of manufacture, or accommodation of utilitarian function or performance. Even if some of the features of the design perform some useful or utilitarian function, this does not mean that the design as a whole is functional since some designs are adopted for the purpose of identification and also to perform a useful function. A particular design that serves some functions should not be found functional unless the design is only one of a limited number of equally efficient options and free competition would be unduly hindered by according that design protection.

This instruction contains errors of two kinds. One, which is not serious, is that being drawn essentially verbatim from judicial decisions on functionality, e.g., Sicilia Di R. Biebow & Co. v. Cox, 732 F.2d 417, 429 (5th Cir.1984), it contains a level of diction inappropriate in an instruction directed to nonlawyers. Words such as “utilitarian” and “optimal” are not found in the vocabulary of all or, we would guess, of most jurors, and should therefore be avoided unless the judge defines them, which he did not do in this case. Also, it is inappropriate to ask a jury whether “free competition would- be unduly hindered,” without making clear what “free competition” means and what it means to hinder it unduly. But these faults of diction or level are easily remedied with a blue pencil which, if the judge was too busy to wield, he could have told Rogers to wield. The use of language drawn from judicial opinions in jury instructions, while generally deplorable, is too common to bar the proposer from objecting to a substantively erroneous instruction.

Second and more serious, the word “utilitarian” is ambiguous. It can mean anything which gives pleasure, which is the approximate sense in which “utility” and “utilitarian” are used by philosophers; or it can mean useful in a sense which excludes beauty, which is the commoner lay meaning of “utilitarian.” Thus in saying that a design is functional if “dictated by utilitarian characteristics,” or if “superior or optimal in terms of engineering, economy of manufacture, or accommodation of utilitarian function or performance,” the proposed instruction would invite the jury to find the hexagonal end panels of the Rogers tray to be functional unless a hexagon shape is indispensable to the basic function of an office tray, which is to hold letters and other papers; or, what seems quite out of the question as a matter of fact, unless it is cheaper to manufacture than a rectangle. If this suggestion were right, the first manufacturer of office trays — who we may assume put a rectangular end panel on his tray — would have been entitled to trademark that panel, forcing competitors to use hexagons or octagons or rhomboids or whatever, since within very broad limits the shape of the end panel, unlike the circular shape of the wheel or the oval shape of the football, cf. In re Deister Concentrator Co., 289 F.2d 496 (C.C.P.A.1961), is irrelevant to the cost of manufacturing the tray, or to its function if aesthetic considerations are deemed unrelated to function. Cf. Fisher Stoves, Inc. v. All Nighter Stove Works, Inc., 626 F.2d 193 (1st Cir. 1980); Black & Decker Mfg. Co. v. Ever-Ready Appliance Mfg. Co., 518 F.Supp. 607, 617 (E.D.Mo.1981), aff'd, 684 F.2d 546 (8th Cir.1982). And yet as a practical matter the clean lines of a rectangular tray may have considerable aesthetic appeal, in which event allowing it to be trademarked would give the manufacturer more protection than necessary to enable satisfied consumers to identify his tray so that they could be sure of continuing to buy from him; and perhaps it would not be possible to give the manufacturer protection in just a rectangle of given dimensions, see Keebler Co. v. Rovira Biscuit Corp., 624 F.2d 366, 378 (1st Cir.1980), because consumers would not be able to distinguish it from competing manufacturers’ rectangles.

As we noted in Chase v. Consolidated Foods Corp., 744 F.2d 566, 570 (7th Cir.1984), the question of what the appellate court should do when the trial judge gave an erroneous instruction to which a party objected, but that party’s proposed instruction was also erroneous, is a vexing one. We do not want to order new trials, at great expense to the parties and the judicial system, because of an error that the losing party’s counsel might have avoided by proposing an instruction that the judge would have realized was more correct than the one he was minded to give. But we also do not think that a federal trial judge discharges his judicial function by taking the position that if both parties tender erroneous instructions on an issue, he can give either one; or worse, that he can ignore a well-founded objection to his own instruction, provided the suggested substitute is also erroneous. Even in an adversarial system of justice the judge is not just an umpire. He has a duty to achieve a just outcome insofar as it is within his power. Thus, “where neither party submits a correct instruction the trial judge’s duty of trial management is not discharged simply by adopting the one that is nearer to the mark.” Exxon Corp. v. Exxene Corp., 696 F.2d 544, 549 (7th Cir. 1982).

We do not for a moment suggest that the district judge disclaimed such a responsibility in this case. He gave the instruction he gave because he thought it was correct, not because he thought that since the plaintiff’s instruction was incorrect, anything went. Moreover, he was not just picking and choosing among the alternative instructions tendered by the parties. Although the instruction that the judge gave adopted the defendant’s general view of functionality, the judge did not give the defendant’s proposed instruction, but wrote his own. And it is possible that if the plaintiff had proposed a correct instruction, the judge would have given that instruction. But we think not. The transcript of the instructions conference does indicate that the judge disliked the word “utilitarian,” and rightly so; but Rogers’ counsel offered to substitute “useful,” without avail. The judge was determined to give his own instruction. It seems quite unlikely that any modification of the plaintiff’s proposed instruction to remove the features we have deemed objectionable would have persuaded the judge to give that instruction instead of his own. There thus is no causal connection between the error in the plaintiff’s proposed instruction and the outcome of the trial, and thus that error is not a good reason for refusing to grant a new trial; so we need not decide which of the instructions — the one the judge gave, or the one the plaintiff proposed — was less erroneous, the approach suggested in Chase for resolving the problem of dual error. See also General Foods Corp. v. Valley Lea Dairies, Inc., 771 F.2d 1098, 1104 (7th Cir.1985) (dissenting opinion).

In holding that there must be a new trial on the issue of functionality, we do not attempt to predict the outcome of that trial. Even if the jury finds that Rogers’ hexagonal end panel design is nonfunctional, it will have to go on and consider the other elements of common law trademark infringement that Rogers must prove. For future reference in this and other cases we suggest that the judge direct the jury to answer all the other questions even if it again answers “no” on whether the design is nonfunctional. There are two reasons for this suggestion. First, a jury should so far as possible not be instructed in a way that makes it much easier to decide in favor of one party than in favor of the other. To decide for Keene it had only to say “no” to question one; to decide in favor of Rogers it would have had to answer more questions, besides assessing damages. (This of course is an argument for the “bifurcated” mode of trial, in which liability and damages are tried separately.) Second, if the jury had been directed to answer all the questions in the special verdict, it might — by answering “no” to another — have mooted any error in the instruction on question one. Maybe Keene is right and Rogers is bound to lose on one of the other questions, in which event our decision of this appeal will delay but not change the outcome of the case, and merely put the parties and the district court to the bother of a second trial. We would know for sure if the jury had been directed to answer the subsequent questions.

For further guidance on remand, we shall discuss briefly what we think a proper instruction on functionality should contain, without for a moment meaning to frame such an instruction. We are writing for judges and lawyers, not for jurors. It is the district judge’s job to put our thoughts into words that will communicate effectively to lay persons. To summarize our earlier discussion of functionality, the jury has to determine whether the feature for which trademark protection is sought is something that other producers of the product in question would have to have as part of the product in order to be able to compete effectively in the market— in other words, in order to give consumers the benefits of a competitive market — or whether it is the kind of merely incidental feature which gives the brand some individual distinction but which producers of competing brands can readily do without. A feature can be functional not only because it helps the product achieve the objective for which the product would be valued by a person indifferent to matters of taste, charm, elegance, and beauty, but also because it makes the product more pleasing to people not indifferent to such things. But the fact that people like the feature does not by itself prevent the manufacturer from being able to use it as his trademark. He is prevented only if the feature is functional, as defined above, that is, only if without it other producers of the product could not compete effectively. If it is nonfunctional, it can be trademarked even though it is pleasing. And a feature is not functional merely because, if someone happens to own an item that has the feature, he might want any other item displayed or connected with it to have the identical fea ture so that the two items would look alike, would be a matched pair.

Since the case must be retried we shall resolve the other issue tendered by Rogers, which is whether the judge should have admitted, as he did, evidence introduced by Keene to show that some consumers buy the Rogers tray because of its appearance rather than because it is made by Rogers or is the Rogers tray. Rogers objected to this evidence because a trademark’s validity does not depend on consumers’ actually liking the manufacturer or even knowing who the manufacturer is. No doubt many people buy “Tide” without knowing that it is made by Procter & Gamble. But as we said earlier the purpose of a trademark is to identify a particular producer or product; it is not essential that consumers know who the producer is. With a name, functionality is rarely an issue. With a slogan, however, it often is. See, e.g., Damn I’m Good, Inc. v. Sakowitz, Inc., 514 F.Supp. 1357, 1362 (S.D.N.Y. 1981). And with a shape, too, it often is, and consumer reactions are probative. If many consumers consider the hexagonal end panel not the distinctive sign of a particular brand of tray (whether or not they know that Rogers makes the brand) but a design feature enhancing the attractiveness, value, and hence utility in the economic sense (as distinct from the sense used in the plaintiff's proposed instruction, where “utilitarian” means practical, no-nonsense, unadorned), this is evidence bearing on the functionality of the feature. Rogers presented evidence that it adopted the hexagonal shape not because it thought a hexagonal shape more attractive than a rectangular one but to distinguish its tray from trays made by its competitors; Keene’s evidence was some contrary evidence. It may not have been very systematic but neither was the survey evidence put in by Rogers to show that consumers recognize the Rogers tray as such. As a matter of fact the two bodies of evidence were pretty consistent, since Rogers’ survey showed that only 24 percent of the respondents (sales persons at midwestern office supply companies) recognized the Rogers tray.

We understand Rogers’ indignation at the thought that its trademark might be jeopardized by the fact that consumers like the trademark for itself rather than just for its informational content (in identifying the product or the producer). However, Rogers’ situation, should it eventually lose the case, would be no different from that of the producer of a brand that is so popular that the brand name has become the generic name (e.g., aspirin, brassiere, cellophane, dry ice, escalator, shredded wheat, yo-yo), and has thereby lost the protection of trademark law. Because trademarks do not have fixed time limits like copyrights and patents, other and vaguer methods are used to cut them off at the point where their value as information about product origin is exceeded by their cost in impeding competition. It is just another example of how firms that have the good fortune to succeed may find themselves put under restrictions that their less fortunate fellows are free from; but the successful firms have at least the consolation of success. In the case of generic names, trademark protection would impose on competitors undue costs of identifying their product, which consumers have come to know by the brand name. In the case of design features that are functional in the aesthetic sense (even more clearly in the “utilitarian” sense), trademark protection would impose undue costs (which will often be greater than the added information costs of not being able to use a brand name that has become generic) of engineering around a feature that the consumer has come to value for its own sake rather than for its role as an identifier of origin. Though a producer does not lose a design trademark just because the public finds it pleasing, there may come a point where the design feature is so important to the value of the product to consumers that continued trademark protection would deprive them of competitive alternatives; and at that point the protection ceases. It is not a disaster for the owner of the trademark; should Rogers lose this case in the end and have to share its hexagonal design with competitors, it can (as it has begun to do) imprint a verbal or other trademark more emphatically, in order to identify its brand to consumers who can no longer look to the hexagon for identification of source. What Rogers may fear of course is not the loss of an identifying mark but the loss of a competitive advantage stemming from the exclusive possession of a popular design; but to protect the intellectual property that consists not of an identifying mark but of a pleasing design a manufacturer must seek the aid of the design-patent law, with its stringent requirements and its 14-year limitation, and not the aid of the trademark laws.

Reversed and Remanded.

[[Image here]]