POSNER, Chief Judge.

Both parties, PIL and Landoll, are publishers of cookbooks and children’s books. PIL, the plaintiff, contends that its books have a distinctive “trade dress” which Lan-doll has copied in violation of section 43(a)(1) of the Lanham Act, 15 U.S.C. § 1125(a)(1), and of parallel state statutes unnecessary to discuss separately. The district judge granted summary judgment for Landoll.

The term “trade dress” refers to the appearance of a product when that appearance is used to identify the producer. To function as an identifier, the appearance must be distinctive by reason of the shape or color or texture or other visible or otherwise palpable feature of the product or its packaging. If it isn’t distinctive, it won’t be associated in the mind of the consumer with a specific producer. If it is distinctive, and if as a result it comes to identify the producer, the danger arises that the duplication of this appearance, this “trade dress,” by a competing seller will confuse the consumer regarding the origin of the product; the consumer may think it the product of the producer whose trade dress was copied. Trade dress thus serves the same function as trademark, and is treated the same way by the Lanham Act and the cases interpreting it. Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 773, 112 S.Ct. 2753, 120 L.Ed.2d 615 (1992); Blau Plumbing, Inc. v. S.O.S. Fix-It, Inc., 781 F.2d 604, 608 (7th Cir.1986); 1 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 8:1, pp. 8-3 to 8-4 (4th ed.1998).

The courts have struggled to articulate a standard for when a trade dress is sufficiently distinctive to be entitled to the prima facie protection of the Lanham Act. See, e.g., Dunn v. Gull, 990 F.2d 348, 351 (7th Cir.1993); Insty Bit, Inc. v. Poly-Tech Industries, Inc., 95 F.3d 663, 672 (8th Cir.1996); Duraco Products, Inc. v. Joy Plastic Enterprises, Ltd., 40 F.3d 1431, 1448-51 (3d Cir.1994); Knitwaves, Inc. v. Lollytogs Ltd., 71 F.3d 996, 1008-09 (2d Cir.1995); Seabrook *339 Foods, Inc. v. Bar-Well Foods Ltd., 568 F.2d 1342, 1344 (C.C.P.A.1977). This struggle is mirrored in the district judge’s opinion and in the briefs of the parties. But efforts to define intuitive concepts such as “distinctiveness” are often both futile and unnecessary. We use with perfect clarity many words that we can’t define, such as “time,” “number,” “beauty,” and “law.” Everyone can recognize when a product has a “distinctive” appearance, without having been tutored in the meaning of “distinctiveness.” But “beyond stating the principles to be applied there is little to be said except to compare the impression made by the two” trade dresses. Joseph Schlitz Brewing Co. v. Houston Ice & Brewing Co., 250 U.S. 28, 30, 39 S.Ct. 401, 63 L.Ed. 822 (1919) (Holmes, J.). So let us consider the principles, or considerations, applicable to the decision. On the one hand, a seller should be encouraged to make his products recognizable by consumers at a glance as his product and not that of another seller. This way the seller will be able to appropriate the benefits of making a product that consumers like, and so he will have an incentive to make a good product. Other sellers won’t be able to free ride on his efforts by tricking the consumer into buying an inferior product from them in the belief that it is the product of the seller whom they have grounds to trust.

On the other hand, a seller should not be allowed to obtain in the name of trade dress a monopoly over the elements of a product’s appearance that either are not associated with a particular producer or that have value to consumers that is independent of identification. In the lingo of unfair competition, elements of the latter type — elements whose value is not merely signification — are a product’s “functional” features; They can be either utilitarian in the narrow sense of that word, or aesthetic. Qualitex Co. v. Jacobson Products Co., 514 U.S. 159, 164-65, 115 S.Ct. 1300, 131 L.Ed.2d 248 (1995); Thomas & Betts Corp. v. Panduit Corp., 138 F.3d 277, 297-98 (7th Cir.1998); W.T. Rogers Co. v. Keene, 778 F.2d 334, 338-40, 343 (7th Cir.1985). Wine is sold corked. The cork is a functional feature of the product, because it enables the wine to age properly; and so the first seller of wine could not claim that the cork was his trade dress. Mink coats are normally sold dyed. The dye does not make the coat any warmer, but it makes it more beautiful, and, once again, it could not be claimed as trade dress by the first furrier to have hit on the idea. Functional improvements may be patentable, or protected as trade secrets, but they cannot be appropriated in the name of trade dress even if they are distinctive.

It would be arbitrary as well as puritanical and even philistine to deny that one function of modern consumer packaging is to be beautiful, the motivation being sometimes a hope that the consumer will infer the quality of the product from the beauty of the package and sometimes a hope that the consumer will derive utility (and so be willing to pay more) from the packaging directly, as when a consumer displays a shapely bottle of champagne to his dinner guests. A producer cannot in the name of trade dress prevent his competitors from making their products as visually entrancing as his own. Ordinarily there is a sufficient variety of pleasing shapes, sizes, colors, and ornamentation to enable beauty without sacrificing differentiation. But if consumers derive a value from the fact that a product looks a certain way that is distinct from the value of knowing at a glance who made it, then it is a nonappropriable feature of the product.

So there cannot be an objection in principle to the concept of aesthetic functionality as a limitation on the legal protection of trade dress. And notice how the “utilitarian” and the “aesthetic” merge in the case put earlier of corking. The cork improves the taste of the wine; the dye enhances the beauty of the mink coat; taste is akin to beauty. But the critics who argue that in application the concept is mischievously vague certainly have a point, see, e.g., 1 McCarthy, supra, §§ 7:80, 7:81, though not one necessary to labor further here.

Formally, distinctiveness and functionality are separate issues. While the burden of proving distinctiveness is of course on the plaintiff, some courts, including our own, hold that functionality is an affirmative defense and so the burden of proof rests on the *340 defendant. E.g., Thomas & Betts Corp. v. Panduit Corp., supra, 138 F.3d at 297; Knitwaves, Inc. v. Lollytogs Ltd., supra, 71 F.3d at 1006. Other courts place the burden on the plaintiff. E.g., Versa Products Co. v. Bifold Co., 50 F.3d 189, 199 (3d Cir.1995); Aromatique, Inc. v. Gold Seal, Inc., 28 F.3d 863, 869 (8th Cir.1994). Others have passed. E.g., Tools USA & Equipment Co. v. Champ Frame Straightening Equip., Inc., 87 F.3d 654, 657 n. 2 (4th Cir.1996); TEC Engineering Corp. v. Budget Molders Supply, Inc., 82 F.3d 542, 546 n. 3 (1st Cir.1996). The uncertainty is limited to unregistered trademarks or trade dress (PIL’s trade dress is unregistered); registration creates a presumption of validity, implying that the defendant has the laboring oar on all issues relating to validity, including functionality, as held in Aromatique, Inc. v. Gold Seal, Inc., supra, 28 F.3d at 869.

Very little turns on resolving the circuit conflict, both because evidence of functionality is equally available to both parties and because functionality and distinctiveness are intertwined issues. Functional features are by definition those likely to be shared by different producers of the same product and therefore are unlikely to identify a particular producer. The principal exception is for new products. If a product is truly new, its functional features may be new too and so may help distinguish it from other products and thus identify the producer. Nevertheless, these features cannot be appropriated; otherwise, competitors would be' prevented from duplicating the new product even to the extent permitted by the branches of the law of intellectual property that protect innovation rather than designations of source. The first producer of an automobile designed a product that had a distinctive appearance; but if its distinctiveness was due to the fact that it had an engine in front instead of horses, and a crank, radiator grill, and steering wheel, the producer would not be able to claim that this combination of functional features constituted trade dress; for that would give him a monopoly of the production of automobiles, and trademark law is not intended to confer product monopolies. W.T. Rogers Co. v. Keene, supra, 778 F.2d at 339.

There isn’t much more to be said at the level of doctrine about the issue in this case, except that a grant of summary judgment on the basis either that the plaintiffs trade dress is not distinctive or that its distinctiveness rests on functional features of the dress is proper only if no reasonable trier of fact could conclude otherwise. See, e.g., Thomas & Betts Corp. v. Panduit Corp., supra, 138 F.3d at 297-300; Sands, Taylor & Wood Co. v. Quaker Oats Co., 978 F.2d 947, 952 (7th Cir.1992); OddzOn Products, Inc. v. Just Toys, Inc., 122 F.3d 1396, 1408 (Fed.Cir.1997). For while it is true that distinctiveness is not really an issue of fact, but rather an issue of the application of a legal standard to facts, it is classified with issues of fact for purposes of drawing the line between the jury’s authority and that of the judge and the reviewing court. Disc Golf Ass’n, Inc. v. Champion Discs, Inc., 158 F.3d 1002, 1006 (9th Cir.1998); Pebble Beach Co. v. Tour 18 I Ltd., 155 F.3d 526, 537 (5th Cir.1998); FunDamental Too, Ltd. v. Gemmy Industries Corp., 111 F.3d 993, 1001-02 (2d Cir.1997); Epic Metals Corp. v. Souliere, 99 F.3d 1034, 1037 (11th Cir.1996); cf. Mucha v. King, 792 F.2d 602, 604-06 (7th Cir.1986).

The trade dress of PIL’s cookbooks is illustrated in the photograph (unfortunately not in color) that we have captioned “PIL.”

*341 [[Image here]]

The components of the trade dress are the size of the pages (8& inches by 11 inches), the gilded edges of the pages, and the covers, which are oilcloth or the equivalent. Each of these features is functional. The size of the pages is a convenience to the cook, who will *342 usually want to lay the book flat on the kitchen counter (and that’s easier to do if the book is large) and glance at it while cooking. That is easier to do if the print is large — and the larger the print, the larger the page must be if the total number of pages is not to be excessive. When the book is open, moreover, the recipe is on one page and a photograph of the dish that faithful compliance with the recipe is supposed to produce occupies the entire facing page. The larger the page and so the larger the picture, the more inviting the dish looks and so the more likely the reader is to try the recipe. The oilcloth cover makes it easier to keep the book clean, which is an issue with cookbooks because they are quite likely to get spattered in the course of cooking. And the gilded pages are necessary to prevent the unsightly appearance that printers call “bleeding.” If a color picture on a page runs all the way to the page’s edge, then when the book is closed, the side opposite the spine (that is, the side consisting of the ends of the pages) will present a motley appearance. Gilding the edges eliminates the differences in the colors of the page ends. Aesthetic considerations are not alien to book design. Although the adage is that you shouldn’t judge a book by its cover, many people do this; otherwise, many books would be sold without covers, as used to be a common practice in Europe. Some people buy books mainly as wall decoration, and bookbinding is regarded as an art and not just as a trade.

What is true is that the problem of bleeding could be solved by giving the ends of the pages any uniform color; it needn’t be gold. But the color gold on a product or its package is a prime example of aesthetic functionality. This is the case mentioned in Qualitex Co. v. Jacobson Products Co,, supra, 514 U.S. at 165, 115 S.Ct. 1300, where “color plays an important role (unrelated to source identification) in making a product more desirable.” Gold connotes opulence, and so is a standard element of the décor of food products, such as chocolate, that are valued for their rich taste rather than for their nutritional value. It also has a long history of use in bookbinding; the spine of the book in which this opinion is printed is decorated with gilt. Gold is a natural color to use on a fancy cookbook. A different color on PIL’s page ends would have a better claim to be a source signifier; compare a blue container of orange juice with an orange one.

Although none of the functional features of PIL’s cookbooks can be appropriated to serve as a trade dress, it doesn’t follow that the ensemble cannot be. Any product, any package, can be decomposed into elements that are not protectable. August Storck K.G. v. Nabisco, Inc., 59 F.3d 616, 620 (7th Cir.1995). Every distinctive appearance is built up from elements that are not themselves distinctive: an Aubrey Beardsley drawing, for example, is built up of line segments. Cf. Roulo v. Russ Berrie & Co., 886 F.2d 931, 936 (7th Cir.1989). In the case of a consumer product, the elements might all be functional. If the product nevertheless presented a distinctive appearance, that appearance would be eligible for legal protection as trade dress unless it was the only way the product could look, consistent with its performing each of the product’s functions optimally. E.g., Service Ideas Inc. v. Traex Corp., 846 F.2d 1118, 1123 (7th Cir.1988); Fuddruckers, Inc. v. Doc’s B.R. Others, Inc., 826 F.2d 837, 842 (9th Cir.1987); American Greetings Corp. v. Dan-Dee Imports, Inc., 807 F.2d 1136, 1143 (3d Cir.1986); LeSportsac, Inc. v. K Mart Corp., 754 F.2d 71, 76-78 (2d Cir.1985). But obviously the fact that a competitor’s product has the same functional features as the plaintiffs isn’t going to make the latter distinctive; if it didn’t have the same functional features, it might not be the same product.

Every producer of a cookbook has to worry about the cover getting spattered, the cook having difficulty laying the book flat and reading small print from a distance, the photos of the food dishes being too small to entice and inspire, the book as a whole looking cheap, and the pages (when the book is viewed from the side opposite the spine) presenting a ragged appearance because the color pictures that fill them bleed to the end of the page. Because of these concerns that are common to all publishers of fancy cookbooks, it is not surprising that more than one *343 publisher would find it optimal, wholly apart from any desire to confuse consumers about the publisher’s identity, to use the very combination of features that PIL claims compose its trade dress — large pages, an oilcloth cover, gilded page edges. Cookbooks published by different publishers will look rather alike in much the same way that automobiles produced by different automobile manufacturers look rather alike, or like many paperback books designed for the airport trade look rather alike because they have the same page dimensions and the same style of cover — laminated paper with the name of book and author in raised letters. Unabridged dictionaries, board-page books for infants, travel books, encyclopedias, atlases, law books, and comic books are other examples of book genres that contain innumerable look-alikes published by different publishers, look-alikes that do not, however, confuse consumers about the identity of the publisher of any particular book.

We don’t wish to seem wise after the event; someone had to be the first to think up the combination of oilcloth cover, large pages, and gilded edges, and for all we know that was PIL. But to reiterate the theme of this opinion, trademark and trade dress law do not protect originality; they protect signi-fiers of source. Publishers who want to distinguish their look-alikes from then- competitors’ — who want, that is, to make their trade dress identify the publisher' — can easily do so by adopting a distinctive logo, imposing a uniform design, typeface, or color on the covers of their books, printing their name prominently on the covers, or doing all these things, as Penguin does. PIL does none of these things. It’s its own fault if someone casually picking up a cookbook published by PIL and a cookbook published by Landoll thinks they were published by the same company; but it is more likely that the consumer would merely think the two books the same type of cookbook.

We have said nothing as yet about PIL’s children’s books. They are similar to the cookbooks, except that the pages are smaller (8$ by 8$) and the books shorter. The pages have gilded edges and the covers an oilcloth feel. But all three features (page size, gilding, and cover texture) are functional for children’s books as for cookbooks. Children need large print, hence large pages, and they like large pictures more than small ones (and large pictures require large pages); in some children’s books the color pictures would bleed to the margin if the ends of the pages weren’t uniformly colored; children frequently have dirty hands that stain everything they touch; and children love gold. PIL’s children’s books look like generic children’s books rather than like the books of a particular publisher. They do not have a uniform series title (like “Golden Books”), the covers do not have uniform colors or designs, and the publisher’s name is not prominently displayed. They are no more distinctive than the cookbooks.

Because the district judge found, correctly in our judgment, that PIL’s books are not distinctive, he didn’t reach the question whether Landoll’s books are so similar as to be likely to confuse consumers. But as the ultimate issue in a trademark or trade dress case is likelihood of confusion, we were struck by PIL’s failure to include in its appendix to either its main brief in this court, or its reply brief, photographs either of any of its own books or any of the books of Landoll that it claims are infringing its trade dress. In intellectual property cases, a picture, or, better, the very object claimed to infringe, is worth a thousand words of brief. Because Landoll’s books resemble PIL’s quite closely, as shown in the next photograph,

*344 [[Image here]]

PIL might have gotten some forensic mileage out of showing us the publishers’ books juxtaposed. But the resemblance is generic, rather than specific. PIL’s trade dress is *345 not distinctive enough that its look-alikes would fool consumers.

Affirmed.