EASTERBROOK, Chief Judge.

For more than 80 years, Clarin has been making x-frame folding chairs. In 1999 it applied for registration of one particular x-frame design as a trademark. The Patent and Trademark Office issued Registration No. 2,803,875 in January 2004. This is the registered mark:

The principal register describes it thus: "a configuration of a folding chair containing an X-frame profile, a flat channel flanked on each side by rolled edges around the perimeter of the chair, two cross bars with a flat channel and rolled edges at the back bottom of the chair, one cross bar with a flat channel and rolled edges on the front bottom, protruding feet, and a back support, the outer sides of which slant inward."

Clarin was acquired by Greenwich Industries in 1993. We refer to Greenwich Industries as Clarin, the name under which it does business. Harvey Hergott, Clarin's general manager at the time of the acquisition, left later that year and, after honoring a five-year restrictive covenant, joined Specialized Seating, Inc., in 2001, two years after his son Alfred had founded that business. Specialized competes in Clarin's principal market: the sale of folding chairs to auditoriums, sports stadiums, convention centers, and other places that need to deploy lots of seats, which owners want to be as light and compact as possible for storage when not in use. Specialized sells a folding chair that to an untrained eye looks like Clarin's trademark chair. There are differences in construction and detail; the chair is not a slavish knockoff, but the basic design tracks the registered mark. That similarity had led to this litigation in which Specialized sought a declaratory judgment that its design does not violate Clarin's rights under the Lanham Act, 15 U.S.C. §§ 1051- 1129, and Clarin counterclaimed for an injunction.

Clarin has been using the mark long enough for its registration to be "incontestable" under 15 U.S.C. § 1065. The word "incontestable" is misleading, however, because an incontestable mark may be contested, and defeated, on many grounds. The principal effect of incontestable status is to relieve the trademark proprietor of any need to show "secondary meaning" — in other words, whether consumers associate a mark with a particular producer is not relevant when the mark is incontestable. Clarin's customers buy in bulk and are sophisticated; it is exceedingly unlikely that any of them is confused about who makes which chair. Because the mark is incontestable, however, the absence of confusion does not matter. Still, nine defenses are available to a person charged with infringing an incontestable mark. 15 U.S.C. § 1115(b). Specialized Seating asserted two of these defenses: that "the registration . . . was obtained fraudulently" (§ 1115(b)(1)) and that "the mark is functional" (§ 1115(b)(8)).

The district court held a bench trial and ruled in Specialized's favor on both of these issues. 472 F.Supp.2d 999 (N.D.Ill. 2007). The judge found that the x-frame construction is functional because it was designed to be an optimal tradeoff between a chair's weight (and thus its cost, since lighter chairs use less steel) and its strength; an x-frame chair also folds itself naturally when knocked over (an important consideration for large auditoriums, where it is vital that chairs not impede exit if a fire or panic breaks out); the flat channel at the seat's edge, where the attachment to the frame slides so that the chair can fold, was designed for strength and attaching hooks to link a chair with its nearest neighbor; the front and back cross bars contribute strength (and allow thinner tubing to be used in the rest of the frame); and the inward-sloping frame of the back support allows the chair to support greater vertical loads than Clarin's older "a-back" design, which the "b-back" design, depicted in the trademark registration, succeeded. The a-back design is on the left and the b-back on the right:

Clarin chairs with a-back designs failed when the audience at rock concerts, seeking a better view, sat on top of the chairs' backs and put their feet on the seats. The tubing buckled at the bend in the frame. The b-back design is less likely to buckle when someone sits on it, and it also produces a somewhat wider back, which concert promoters see as a benefit. (Patrons sometimes try to get closer to the stage by stepping through rows of chairs. The gap between b-back chairs is smaller, so they are more effective at keeping crowds in place.)

Having concluded not only that the overall design of Clarin's chair is functional, but also that each feature is functional, the district judge added that Clarin had defrauded the Patent and Trademark Office by giving misleadingly incomplete answers to the trademark examiner's questions. The examiner initially turned down Clarin's proposal to register the design as a trademark, observing that the design appeared to be functional. Clarin replied that the design was chosen for aesthetic rather than functional reasons. (This was not a complete answer, as attractiveness is a kind of function. See Jay Franco Sons, Inc. v. Franek, No. 09-2155, 615 F.3d 855, 860-61, 2010 WL 3156539 (7th Cir. Aug. 11, 2010). 10-12. But we need not pursue that subject.) Clarin observed that a patent it held on an x-frame chair, No. 1,943,058, issued in 1934, did not include all of the features in the mark's design. What Clarin did not tell the examiner is that it held three other patents on x-frame designs: No. 1,600,248, issued in 1926; No. 2,137,803, issued in 1938; and No. 3,127,218, issued in 1964. The district judge concluded that the four patents collectively cover every feature of the design submitted for a trademark except the b-back, and that as the b-back is a functional improvement over the a-back Clarin should have disclosed all of these utility patents. Had it done so, the judge thought, the examiner would have refused to register the proposed mark.

Clarin's appeal was delayed by settlement negotiations. Briefing and argument were deferred until after they broke down. It was further delayed by the district court's failure to enter a proper judgment, a common problem in the Northern District of Illinois. See, e.g., Rush University Medical Center v. Leavitt, 535 F.3d 735 (7th Cir. 2008). The judge stated that Specialized is entitled to a declaratory judgment that the registration is invalid because of both fraud and functionality. A declaratory judgment must be set out on a separate document containing its terms. Fed.R.Civ.P. 58(a). The judgment in this case does not do that. The parties, and perhaps the district judge, seem to have assumed that, if the judge's opinion names the winner, no one need bother with the step of producing a concise declaration in a separate document. But "[i]f courts are to require that others follow regular procedures, courts must do so as well." Hollingsworth v. Perry, ___ U.S. ___, 130 S.Ct. 705, 715, ___ L.Ed.2d ___ (2010). Clarin complicated matters by omitting the defective judgment from the appendix to its brief, despite the requirement in Circuit Rule 30(a) that it be included — and despite counsel's representation to the court, see Circuit Rule 30(d), that Rule 30(a) had been followed. The court discovered the problem on its own, and at oral argument we questioned whether appellate jurisdiction exists given the absence of any relief. See, e.g., Horn v. Transcon Lines, Inc., 898 F.2d 589 (7th Cir. 1990); Azeez v. Fairman, 795 F.2d 1296 (7th Cir. 1986). We invited the parties to return to the district court, and ask the judge to issue a proper declaratory judgment. They did, the judge did, and Clarin's new appeal (No. 10-2670) has been consolidated with the original one (No. 07-1435). The case is at last ready for decision.

Findings of fact made after a bench trial must stand unless clearly erroneous. Fed.R.Civ.P. 52(a)(6); Anderson v. Bessemer City, 470 U.S. 564, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985). Although "functionality" is the "ultimate issue" in a case such as this, an "ultimate" issue (such as whether the defendant was negligent or engaged in racial discrimination) remains one of "fact" under this standard, because it is generated by applying legal principles to factual conclusions. See, e.g., Icicle Seafoods, Inc. v. Worthington, 475 U.S. 709, 106 S.Ct. 1527, 89 L.Ed.2d 739 (1986); Pullman-Standard v. Swint, 456 U.S. 273, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982). "Functionality" certainly isn't an issue of law; it represents a fact-specific conclusion about whether aspects of a design are "essential to the use or purpose of the article or if it affects the cost or quality of the article". Inwood Laboratories, Inc. v. Ives Laboratories, Inc., 456 U.S. 844, 850 n. 10, 102 S.Ct. 2182, 72 L.Ed.2d 606 (1982); see also TrafFix Devices, Inc. v. Marketing Displays, Inc., 532 U.S. 23, 32-33, 121 S.Ct. 1255, 149 L.Ed.2d 164 (2001). Many decisions apply Rule 52's standard to a finding that a mark is functional. See, e.g., Service Ideas, Inc. v. Traex Corp., 846 F.2d 1118, 1123 (7th Cir. 1988); Pebble Beach Co. v. Tour 18 I Ltd., 155 F.3d 526, 537 (5th Cir. 1998); Fuji Kogyo Co. v. Pacific Bay International, Inc., 461 F.3d 675, 681 (6th Cir. 2006); Clamp Manufacturing Co. v. Enco Manufacturing Co., 870 F.2d 512, 514 (9th Cir. 1989); Epic Metals Corp. v. Souliere, 99 F.3d 1034, 1037 (11th Cir. 1996).

Clarin tries to evade this standard of review by contending that the district judge's findings were influenced by legal errors. We don't see any. The district judge started from the proposition, which the Supreme Court articulated in TrafFix, that claims in an expired utility patent presumptively are functional. Since utility patents are supposed to be restricted to inventions that have utility, and thus are functional, that's a sensible starting point — and since inventions covered by utility patents pass into the public domain when the patent expires, it is inappropriate to use trademark law to afford extended protection to a patented invention. See also Jay Franco, 615 F.3d at 857-59. 5-8. Clarin itself obtained four utility patents for aspects of the x-frame folding chair. These patents disclose every aspect of the asserted trademark design except for the b-back. And the district judge did not commit a clear error by concluding that the b-back design is a functional improvement over the a-back design. This means that the trademark design is functional as a unit, and that every important aspect of it is independently functional. It looks the way it does in order to be a better chair, not in order to be a better way of identifying who made it (the function of a trademark).

We do not doubt that there are many other available functional designs. Sometimes the function of the functionality doctrine is to prevent firms from appropriating basic: forms (such as the circle) that go into many designs. Our contemporaneous opinion in Jay Franco discusses that aspect of the functionality doctrine. This does not imply that preserving basic elements for the public domain is the doctrine's only role.

Another goal, as TrafFix stressed, is to separate the spheres of patent and trademark law, and to ensure that the term of a patent is not extended beyond the period authorized by the legislature. A design such as Clarin's x-frame chair is functional not because it is the only way to do things, but because it represents one of many solutions to a problem. Clarin tells us that other designs are stronger, or thinner, or less likely to collapse when someone sits on the backrest, or1 lighter and so easier to carry and set up. Granted. But as Clarin's `248 patent states, the x-frame design achieves a favorable strength-to-weight ratio. Plastic chairs are lighter but weaker. Y-frame chairs are stronger but use more metal (and so are heavier and more expensive);, some alternative designs must be made with box-shaped metal pieces to achieve strength, and this adds to weight and the cost of fabrication. The list of alternative designs is very long, and it is easy to see why hundreds of different-looking folding chairs are on the market.

What this says to us is that all of the designs are functional, in the sense that they represent different compromises along the axes of weight, strength, kind of material, ease of setup, ability to connect ("gang") the chairs together for maximum seating density, and so on. A novel or distinctive selection of attributes on these many dimensions can be protected for a time by a utility patent or a design patent, but it cannot be protected forever as one producer's trade dress. When the patent expires, other firms are free to copy the design to the last detail in order to increase competition and drive down the price that consumers pay. See, e.g., Bonito Boats, Inc. v. Thunder Craft Boats, Inc., 489 U.S. 141, 109 S.Ct. 971, 103 L.Ed.2d 118 (1989); Sears, Roebuck Co. v. Stiffel Co., 376 U.S. 225, 84 S.Ct. 784, 11 L.Ed.2d 661 (1964). Once Harvey Hergott's restrictive covenant expired, he was free to do that too.

Clarin reminds us that a product whose overall appearance is distinctive can be protected under the trademark laws, even though most of the product's constituent elements serve some function. See, e.g., Service Ideas, 846 F.2d at 1123; W.T. Rogers Co. v. Keene, 778 F.2d 334, 339-40 (7th Cir. 1985). That's true enough, but what made the appearance "distinctive" in these and similar decisions was a nonfunctional aspect of the design. For example, Keene dealt with a letter tray that had irregular hexagons as end caps; each hexagon had an interior cutout. End caps serve the function of making letter trays stable and separating the lower tray from the upper tray, and cutouts make them lighter while reducing the cost of materials, but neither the irregular shape of the hexagon nor the shape of the cutout served any function. They made the tray distinctive and enabled consumers to determine which firm made it. If Clarin had placed a cutout in the backrest, or given it a distinctive pattern, it could claim those attributes as trade dress. But what Registration No. 2,803,875 claims is the x-frame profile with three cross bars and a slanted back support. All of the claimed features are functional; none was added to produce a distinctive appearance, that would help consumers identify the product's source.

Because the district court did not commit clear error in finding Clarin's design to be functional, it is unnecessary to decide whether Clarin committed fraud on the Patent and Trademark Office. All a finding of fraud does is knock out the mark's "incontestable" status, and its registration, under § 1115(b)(1). It does not affect the mark's validity, because a trademark need not be registered to be enforceable. See Orient Express Trading Co. v. Federated Department Stores, Inc., 842 F.2d 650, 653-54 (2d Cir. 1988) (a trademark can be enforced under 15 U.S.C. § 1125(a) even after its federal registration is cancelled for fraud); Far Out Productions, Inc. v. Oskar, 247 F.3d 986, 996 (9th Cir. 2001) (a trademark can be enforced under state common law even after its federal registration is cancelled for fraud); see also J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 31:60 (2006 ed.). A finding of fraud therefore would not end the case; a finding of functionality does.

Clarin contends that we should consider this topic even though it does not affect Specialized's right to sell a copycat chair, because it might matter to attorneys' fees, but fees have not been awarded. A decision on legal issues just because they might matter to some later dispute would be advisory. Clarin also appears to be concerned that the district judge's finding of fraud might affect future litigation against a different competitor, but issue preclusion (collateral estoppel) applies only to issues actually and necessarily resolved in the first case. Brown v. Felsen, 442 U.S. 127, 139 n. 10, 99 S.Ct. 2205, 60 L.Ed.2d 767 (1979); Restatement (Second) of Judgments § 27 (issue preclusion operates only if the earlier resolution was "essential to the judgment"). It was not necessary to address fraud on the PTO, so the district judge's opinion on this subject does not have preclusive effect.

AFFIRMED