The Sacrificed Books of The Ninth Gate: Inquiry into Polanski’s Cursed Book

25 October 2026 Hugues Stories 0

Dear bibliophiles, good day.

By an adept, in the manner of Lucas Corso, surveyor of the bibliophilic margins.

Some years ago I was told about a man in Toulon who owned Latin leaves of the seventeenth century whose provenance was murky. Not murky in the way ordinary bibliophiles mean it: no dubious trimming of the margins, no ill-judged restoration. Murky because these leaves, he said, came from a book that had been taken apart on purpose to make another book, one that had never existed. I noted the name, filed the information under things to check, and then forgot it. Until the other evening, when I watched Roman Polanski’s The Ninth Gate again.

The film, released in 1999, adapts Arturo Pérez-Reverte’s El Club Dumas. It tells of the quest of a rare-book broker, a trade I know well, commissioned to authenticate one of the three known copies of a book called De Umbrarum Regni Novem Portis, On the Nine Gates of the Kingdom of Shadows, supposedly printed in Venice in 1666 by one Aristide Torchia, and said to have been co-signed by the Devil himself. Torchia, burned by the Inquisition, was supposed to have left three copies identical except for nine engravings which, compared with one another, would yield the keys to invoking Lucifer. The book does not exist, of course. Pérez-Reverte invented it for his novel. Polanski had it manufactured for his film. And this is where the bibliophilic story starts to get interesting.

Three false books, two real ones sacrificed. The production designer of The Ninth Gate is Dean Tavoularis, the man behind The Godfather, Apocalypse Now and Bonnie & Clyde. Oscar, 1975. A man of that stature has a team of prop makers, and the film’s head prop master is called Gilbert Pieri. Pieri had to make three hero copies of the Ninth Gate: the Balkan, the Fargas and the Kessler. Three distinct bindings, nine engravings per volume and, a decisive detail, a Latin text inside that had to look real under the camera, in close-up, open on Johnny Depp’s knees.

Pieri did not print that text. He did what any bibliophile will recognise as heresy: he bought two authentic old Latin books, one of more than five hundred pages, the other of more than eight hundred, and he took them apart. Some twenty leaves from the two volumes were combined, reinserted into the custom-made bindings, and duplicated to serve the three copies needed for the film. The detail is known to specialist collectors, who tell it to one another on forums devoted to film props. Some claim to own original pages from these two books, torn out during production and given to members of the crew. One of them eventually identified the source book by typing a Latin phrase from the film into Google: it was a Latin theological and Catholic work of 1588, from which the production had merely removed a few words to give the illusion of a coded text.

Some twenty leaves of a book of 1588, dismembered to appear in a Polanski film: this should make any lover of old paper tremble. Two real books were immolated on the pretext of making a third, rarer still, since it does not exist.

Then there are the nine engravings. They are, in Pérez-Reverte’s novel as in the film, the central element: each copy contains nine, but three of them differ from one volume to the next, and comparing the eighteen images reveals the secret. These engravings, too, have a bibliophilic history worth attention.

Here is a rare particularity: the engravings already existed in Pérez-Reverte’s novel, where they bear the initials AT and LCF; those in the film take up their formal language. The formal language of these engravings, as everyone knows, draws ostensibly on the Hypnerotomachia Poliphili, the allegorical romance printed by Aldus Manutius in Venice in 1499 and attributed to Francesco Colonna. It is, as everyone knows, one of the most prestigious incunabula ever printed, famous for its one hundred and seventy-two anonymous woodcuts, themselves the object of five centuries of speculation about their author. A copy can be consulted online on Gallica, the digital library of the Bibliothèque nationale de France (bpt6k587910); those who want to see what a perfect copy costs today will remember that a Poliphilo in a fine binding has passed four hundred thousand euros at public sale, and that Jean Grolier alone owned five copies.

To make his cursed book, then, Pérez-Reverte had the formal language of the rarest and most mysterious book of Western printing adapted. It is a choice of great elegance: a seventeenth-century fake imitating a real book of the late fifteenth. The film’s engravings were drawn, engraved and printed like real copperplate prints, nine for each copy, twenty-seven prints in all, not counting the variants that Polanski had redrawn. Those prints exist somewhere, framed at Polanski’s or at Pieri’s, or scattered among the members of the crew.

As for official sales, the record is poorer than for The Name of the Rose. Neither Christie’s, nor Sotheby’s, nor Bonhams, nor Heritage, nor Propstore has seen a hero copy of the Ninth Gate go through. The three bindings made by Pieri are, to my knowledge, still in private collections; one belongs to the collection of a certain “balkan.press”, a discreet figure who runs a specialist website devoted to the object.

But the parallel market is teeming. More than forty official replicas of the book are listed online, sold for between five hundred and twelve hundred dollars depending on quality. A whole tinkerer’s literature circulates, with manufacturing tutorials, reconstructed typefaces, tracings of the nine engravings and of the tenth (the one called the “true ninth gate”, inscribed LCF). The Ninth Gate is probably the most copied prop book in the history of cinema.

And then there are the leaves. Those from the two sacrificed books. They circulate quietly, sometimes on eBay, more often in private transactions between collectors. The bibliophilic status of these objects is staggering: they are at once fragments of real old books (the Latin theological work of 1588), film props deposited by the production, and relics of the shoot co-signed Polanski. Three provenances for a single leaf: for the collector, an object to the third degree.

I do not know whether the man from Toulon really has Latin leaves that passed through the shooting of The Ninth Gate. I did not bother to go and see. But the idea that real old books were destroyed (consumed is the right word) to make a fictional book whose market value reaches several hundred euros today in replica, and incalculably more as a hero copy, should give us pause.

Polanski and his team inverted the bibliophilic equation. Where the usual market pays for the old manuscript and despises the modern copy, their film does the opposite: it buys and destroys a theological book of 1588 to make a modern fake that resells for more than the pages it is woven from. The cursed book has done its work. The real work has dissolved into the matter of the fake. And the leaves now in circulation are, as in the film, fragments that must be compared with one another to guess where they come from.

Burned books vanish, but those that are shut away always end up coming back.

Guild shelfmark: GBO-C/2 — Le Livre Démembré [The Dismembered Book]

Document classified: restricted circulation — Bureau of Litigious Props.

La lettre du bibliophile

Si cette chronique vous a plu : recevez trois nouvelles par semaine — livres rares, reliures, enchères, et les Dossiers IGLI. 18 ans d'archives, quelques milliers de lecteurs.

Pas de pub, pas de revente de données. Désabonnement en un clic.

À lire ensuite

IGLI Clinical Files: The Bibliophile Lost for 24 Hours on a Classifieds Site

Dear bibliophiles, good day.

By Dr. Achille Dandillot, physician-inspector in the clinical section of the IGLI, holder of a chair he created himself and that nobody thought of disputing. File no. 2, reference IGLI-DC-2/207.

The patient’s sister telephoned at ten past seven, and she spoke low. Her brother had gone onto the classifieds site the evening before just to look, and since then, nothing more. He had not dined, he had not slept. He could be heard, from time to time, repeating prices with no sentences around them: ninety euros, a thousand euros, fourteen euros eighty. Around five in the morning he had said aloud, to himself, “this one might be the one.” His sister waited two hours before calling, out of consideration for him.

I record the symptom under the name of price palilalia. The patient no longer repeats words, he repeats prices, and the price is the only element that survives from a listing stripped of its description. We had not yet met this stage in so practiced a subject.

Mr. R. is fifty-two and owns 2,000 volumes, a good part of them in period bindings. He has frequented public sales for twenty years, he reads printers’ Latin, he knows how to collate. Our services already knew him: file IGLI-DC-1/193, gallicanosis in a benign form, resolved without treatment. I add, because the diagnosis depends on it entirely, that he has never in his life bought a Jean de Bonnot.

The place where we found him has no name in our nomenclature. A plain. Or a dump; the word comes to me and I keep it.

Under the two words livre ancien (“old book”), the site’s Books section declares 545,417 listings. The same two words, entered in Antiques, give 2,859, and the first on the list is a marble clock with its urns, €130.

The vocabulary of the site comes down to four adjectives that trade places without anyone losing by it: “old” (ancien), “old” (vieux), “rare,” “collectible.” One therefore finds, on the first screen of every query and without having had to look for them, a Larousse encyclopedia set sold as old books for €10, a lot of rare old books advertised with “over 100 years old” in parentheses for €15, a rare old book of 1862 at €1,000 whose title and author are both withheld, an old book at €8 about which nothing whatever is given, and a rare old book entitled L’Éléphant rose at €14.80. Another listing offers old Victor Hugo books, sold on account of a death, without saying whose, and a pair of Art Deco bookends appears in the same category, where, admittedly, it holds something up. The patient was looking for a book printed before 1700.

The prices have no scale. A Bible in Latin, a Vulgate printed at Lyon in 1732, €90. Three lines higher, a sentimental novel of 1920, €89. Under the word incunable (“incunabulum”), 154 listings mix Gutenberg facsimiles at €100 with a Bartholomaeus Anglicus of 1525 at €2,200; university essays on incunabula appear there too, at €15, with nothing to distinguish them from the objects. Under the word bibliophilie, the very first listing, shown at €8,000, is not a sale: a bookseller is buying there, and he filled in the box because it was compulsory.

The photographs remain. Always the same one: the closed volume, laid on a flowered duvet, shot from three-quarters. Never the title page. Never the spine. The description comes down to two words, “see photos,” and there is nothing to see in the photos. I add, to no purpose, that it is often the same flowers, big pink peonies on a cream ground, and that I ended up wondering which home-linen manufacturer has equipped a whole generation of sellers.

The history had been left open. The patient first searched for an Elzevier and obtained eleven results, four of them for a twentieth-century Dutch scientific publishing house and one for a cat to give away. He then refined, reliure plein veau XVIIe (“17th-century full calf binding”), and received the pair of bookends and an arithmetic textbook of 1883. Precision degrades the result: the engine seems not to know what full calf is, and to retain the listings that share the most letters with the query. He finally typed incunable, the highest word in the vocabulary, and received a Gutenberg facsimile and an essay on woodcut engraving.

He then gave up all querying, typed livre ancien, and stopped reading. I propose to call this state scrolling acedia. Slack jaw, high breathing, thumb descending at regular intervals, loss of the sense of time, critical intelligence intact: questioned the next day, Mr. R. knew perfectly well that the last three hundred listings were worth nothing. We found him at page forty-one, between a lot of Bibliothèque verte (a children’s series) and the arithmetic textbook of 1883, returned there by a path I cannot explain.

We did not speak to the patient and we refrained from telling him he was wasting his time, since he knows it and it is the only thing he still knows. My assistant read aloud, slowly, the complete entry for one book. A real one. Full title, place, printer, date, format, collation, pagination, condition of the binding, nature of the endpapers, accidents, provenances with the names of their bearers. After two minutes, Mr. R. looked up.

The protocol stops there. The cure by catalogue entry, a heavy variant of the gentle bibliographic electrotherapy administered in the previous file, takes three minutes of exact description and requires no apparatus. One does not get a bibliophile off the site by showing him the door; one gets him off by making him hear a described book. The patient asked the time, took a long while to admit that a whole day was missing, and said he had found nothing.

Neither greed nor laziness holds a man that deep in the listings. Hope holds him. Among 545,417 badly named and undescribed listings, there is the 17th-century full calf sold for forty euros by someone who does not know what he holds; I had one in my hands a long time ago, picked up off a sidewalk by a colleague who knew nothing about it and who showed it to me that same evening, delighted. The patient is therefore right, and a man who is right does not get treated. We have no long-term treatment to offer.

The bookseller’s catalogue entry serves first to make it possible to say no. Its collation and its list of defects make the day finishable, and a catalogue makes a passion habitable. The site never delivers its verdict. A man who never obtains a verdict stops only from exhaustion, and exhaustion came, in Mr. R., at the twenty-fourth hour.

Prognosis guarded. The patient slept eleven hours and resumed his work. For practiced subjects, I recommend that a third party call the subject by name, aloud, in the room, after forty minutes of presence on the site, and by the surname, not the first name, the first name waking no one. The other measures I had planned to propose do not stand up against hope, and I withdraw them.

It remains to be seen whether the site has anything to do with it. The Institute draws up no indictment and I impute no intention to a company I do not know, but I note the state of the boxes. A seller must choose a category, state a price and post a photograph.

He may leave the description empty, and many do. No field demands a printing date, a format, a binding condition, whereas the patient never bought anything else than that. The same category takes in a marble clock, a pair of bookends, a lot of Bibliothèque verte and a Bartholomaeus Anglicus of 1525.

The engine appears to rank by shared letters, and the company appears to sell visibility options that can carry listings that say nothing to the head of the list. The price box cannot be left empty.

I looked at the site for twenty minutes on my way home, for the needs of the file. I bought nothing. His sister writes that he no longer wants to hear anything about anything.

He went back Thursday. So did I.

GUILDE · AD/BONCOIN

— Continuer la balade —

Be the first to comment

Leave a Reply

Your email address will not be published.