How to Steal a Million

Snapshot

A Parisian confection about forgery, staged by one of Hollywood's great realists. Nicole Bonnet (Audrey Hepburn) discovers that her father Charles (Hugh Griffith) — a genial, unrepentant forger of Van Goghs and Cézannes — has lent the family's "Cellini" Venus, itself carved by her grandfather, to a private museum, where a routine insurance appraisal will expose the whole dynasty. To recover the statue she enlists a man she believes to be a burglar, Simon Dermott (Peter O'Toole), whom she has already shot in the arm during a botched break-in at her own house. He is in fact an investigator who specializes in art fraud, which makes the caper a double con and the romance a negotiation between two people lying about their professions.

Released by Twentieth Century-Fox in the summer of 1966, the film is Wyler's last comedy and his second and final collaboration with Hepburn, thirteen years after Roman Holiday (1953) made her a star and won her an Academy Award. It arrives at the tail end of a specific commercial ecology — the American studio picture shot in Europe, in anamorphic widescreen and color, built around couture, tourism, and star chemistry — and at the precise moment that ecology began to be swept away.

Industry & production

The production sits inside Fox's post-Cleopatra recovery. The 1963 Taylor-Burton epic had nearly destroyed the studio; The Sound of Music (1965) restored it; the mid-decade program that followed leaned on prestige directors, international locations, and roadshow-adjacent gloss. How to Steal a Million is a modest entry in that program: no spectacle, but every dollar visible on screen in the form of Paris, Panavision, and Givenchy.

Fred Kohlmar produced. Harry Kurnitz wrote the screenplay from a story by George Bradshaw (commonly cited under the title "Venus Rising"). Kurnitz was a house wit of the period — his credits include the Billy Wilder–directed Witness for the Prosecution (1957) — and his dialogue here works in the mode of light epigram rather than farce mechanics, which suits Wyler's unhurried tempo and also accounts for some of the film's slackness.

Shooting took place in and around Paris, with interiors at the Studios de Boulogne. The choice was economic as much as aesthetic: through the 1960s, so-called runaway production let American companies exploit favorable exchange rates, cheaper below-the-line labor, and frozen overseas earnings, while delivering the European settings that widescreen color was thought to require. The film belongs to that logic completely. Its Paris is a location and a commodity at once.

Casting reflects Wyler's habits and the industry's transatlantic mixing. Hugh Griffith had won a Supporting Actor Oscar under Wyler for Ben-Hur (1959). Eli Wallach plays the American collector Davis Leland, an acquisitive innocent whose desire for the Venus and for Nicole are indistinguishable. Charles Boyer appears as an art dealer, and the supporting bench draws on the French industry — including Marcel Dalio, whose presence carries the memory of Renoir's La Règle du jeu (1939) into a later film about collections, games, and social performance. Peter O'Toole, four years past Lawrence of Arabia (1962), was deliberately lightening his register; he had already tested comedy in What's New Pussycat? (1965).

The film's precise commercial performance is not something I will quantify here; the reliable public record is thin, and contemporary trade framing treated it as a respectable rather than outsized performer in a year whose big earners lay elsewhere.

Technology

Photographed in Panavision anamorphic 35mm with color by DeLuxe, projected at roughly 2.35:1. By 1966 anamorphic had matured past the distortions of early CinemaScope: Panavision's optics allowed closer focus, faster lenses, and less of the "scope" barrel-and-stretch that had pushed directors toward friezelike staging in the 1950s. The film exploits that maturity mainly in service of two-shots — the format's real gift to romantic comedy is the ability to hold two faces at conversational distance without cutting.

The plot's own technology is period-specific and worth noting as content rather than craft. The museum's security is an electronic alarm system with sensor coverage and a monitoring room, and the caper turns on defeating it not by superior gadgetry but by exhausting it: trip it enough times with a magnet and a boomerang and a length of string, and the humans in charge will switch it off themselves. This is a joke about the mid-1960s faith in electronic security, and it belongs to a small cycle of films fascinated by alarm systems as dramatic objects — Jules Dassin's Rififi (1955) and Topkapi (1964) most directly. Where the Bond films of the same years fetishized apparatus, Wyler's film sides with improvisation and domestic props.

Technique

Cinematography

Charles Lang shot the film — one of the most decorated cameramen in Hollywood history, with an Oscar for A Farewell to Arms (1933) and a run of nominations spanning four decades. Lang's presence is the strongest formal link between this film and Hepburn's earlier work: he photographed Sabrina (1954) and, for Stanley Donen, Charade (1963), the picture How to Steal a Million most obviously answers.

The style is high-key, clean, and unfussy in daylight, with a marked shift for the museum sequences, where the light narrows to practicals, torch beams, and the glow of vitrines. Lang's handling of Hepburn is worth attention on its own terms: the lighting is designed around the geometry of her face and around Givenchy's silhouettes, which frequently means keeping her in three-quarter figure rather than close-up so the clothes read. The film is, in a fairly literal sense, photographed to be a fashion document as well as a narrative.

Editing

Robert Swink cut the film, as he cut nearly all of Wyler's later work — Roman Holiday, Friendly Persuasion (1956), The Big Country (1958), The Children's Hour (1961), The Collector (1965), and afterward Funny Girl (1968). Wyler's method, established in the 1930s and 40s, was to shoot exhaustively and stage in depth, which produces long takes and comparatively few cuts within a scene. The consequence here is a comedy with an unusual amount of dead air by modern standards: the film runs a little over two hours, and the closet sequence in particular is allowed to expand well past its narrative function. Whether this reads as patience or as slackness is the central critical question about the picture, and both readings have been defended.

Mise-en-scène / staging

Alexandre Trauner designed the production — a fact that should reorganize how the film is seen. Trauner was the designer of Carné and Prévert's poetic realism (Le Jour se lève, Les Enfants du paradis) and later of Billy Wilder's The Apartment (1960), Irma la Douce (1963), and Witness for the Prosecution. His Bonnet mansion is a machine for concealment: a house whose visible function is display and whose actual function is fabrication, with the forger's studio tucked behind the salon. The museum, likewise, is built as a set of transparent enclosures — glass, plinths, sightlines — in which the Venus is simultaneously perfectly visible and perfectly unreachable.

Wyler stages against this architecture in the manner André Bazin praised in his 1948 essay on the director, "William Wyler, or the Jansenist of mise en scène": foreground and background held in the same shot, dramatic relations expressed by placement rather than by cutting. The broom-closet scenes invert the principle — two people jammed into a space too small for the frame, the anamorphic width filled by proximity rather than depth. It is the most formally interesting passage in the film, and its comedy is spatial before it is verbal.

Sound

The score is by John Williams, then early in a film career largely spent on light comedies and television and billed in this period on a number of assignments as "Johnny Williams." The music is jazz-inflected, small-combo, and gently ironic — closer to Henry Mancini's contemporaneous idiom than to anything Williams would become known for after 1975. It works chiefly as tonal insurance: it tells the audience that no real jeopardy is in play. The museum sequences drop the score almost entirely in favor of alarm bells, footsteps, and silence, and the contrast is the film's most effective sound gesture.

Performance

Hepburn plays Nicole as a woman perpetually managing a crisis she did not create, and the performance is calibrated to fatigue as much as charm — the exasperation of a daughter who has been the responsible adult for years. O'Toole plays Simon with a slight abstraction, as if amused by a scheme he already understands better than anyone else in it; the character's secret occupation is legible in the performance well before the plot confirms it. Griffith's Bonnet is the film's engine of moral comedy, an artist who genuinely cannot see the difference between painting a Van Gogh and being one. Wallach supplies a broader, hungrier American energy against which the European characters define themselves.

Narrative & dramatic mode

The structure is a double-deception romance grafted onto a heist. Each protagonist withholds a professional identity from the other; the caper functions as a courtship ritual in which trust is built precisely through cooperative lying. Dramatically, the heist is not the climax but the crucible — it forces sustained physical intimacy on two people who cannot yet be honest, and the confession that follows is the real third act.

The film is notably low-stakes by design. Nothing is at risk except reputation, and the moral scheme quietly exempts the forger from consequence: Bonnet's fakes are permitted to remain in circulation because they give pleasure, and because the collectors who buy them are figures of appetite rather than discernment. This is a comedy of manners in caper clothing, and its dramatic mode is closer to Lubitsch's Trouble in Paradise (1932) — thieves who love each other, property treated as a joke — than to the procedural rigor of Rififi.

Genre & cycle

How to Steal a Million belongs to the mid-1960s glamour-caper cycle: Topkapi (1964), Gambit (1966), and, at the cycle's stylish end, The Thomas Crown Affair (1968). Its more immediate model is Charade (1963), which paired Hepburn with Cary Grant, a Paris setting, a Mancini score, and an identity-shifting leading man — a film this one is plainly designed to rhyme with, down to the shared cinematographer.

It also sits at the end of a Hepburn subgenre: the Paris-and-couture romance running from Funny Face and Love in the Afternoon (both 1957) through Paris When It Sizzles (1964) and Charade. Hepburn's next film, Donen's Two for the Road (1967), takes that persona apart deliberately; Wait Until Dark (1967) abandons it entirely, and a long withdrawal from screen work followed.

Authorship & method

Wyler is the paradigmatic "invisible" studio director, and the critical fight over him is instructive. Bazin treated his restraint as a rigorous ethic, an ascetic refusal of directorial display that placed interpretive freedom with the viewer. Andrew Sarris, in The American Cinema (1968), consigned him to the category "Less Than Meets the Eye" — competent, adaptable, without a governing vision. Wyler's own working method, notorious in the industry, was volume: repeated takes until performances arrived where he wanted them, a practice that earned him the nickname "40-take Wyler."

That method is legible here in the acting, which is unusually settled for a comedy, and in the film's general refusal of pace. It is worth noting what Wyler had just made: The Collector (1965), a claustrophobic two-hander about abduction. How to Steal a Million is an emphatic change of register from a director in his sixties choosing lightness. Two films remained: Funny Girl (1968) and The Liberation of L.B. Jones (1970).

The key collaborators — Lang, Swink, Trauner, Kurnitz, Williams, and Hubert de Givenchy, who had dressed Hepburn since Sabrina — represent an assembly of very high studio craft. The authorship question the film poses is whether that much craft, applied to material this slight, constitutes a style or merely a finish.

Movement / national cinema

Formally American, geographically French, the film is a clean specimen of the transatlantic studio picture. It is not a work of any French movement — the Nouvelle Vague was by 1966 well into its second phase, and nothing in Wyler's approach engages it — but it borrows French labor and French iconography extensively, and Trauner's design in particular imports a genuine lineage of French studio filmmaking into a Fox production. The result is a Paris that is neither documentary nor entirely fantasy: a designed city, legible to American audiences, populated by French character actors carrying histories from an earlier national cinema.

Era / period

1966 is a hinge year. The Production Code was in visible collapse; Who's Afraid of Virginia Woolf? and Blow-Up forced revisions that would culminate in the 1968 ratings system. Bonnie and Clyde and The Graduate arrived the following year. Against that, How to Steal a Million is an artifact of the order about to end: genteel, chaste, expensively appointed, addressed to an adult audience that assumed sophistication meant tailoring and restraint rather than transgression.

The film's subject also has a real-world spine. The Dutch forger Han van Meegeren, whose fake Vermeers fooled the Nazi collector Hermann Göring and led to a celebrated 1947 trial, is the obvious historical precedent for Charles Bonnet, and the 1960s art market's rapid inflation gave the premise contemporary charge. The film precedes the public flowering of the Elmyr de Hory affair, which Orson Welles would turn into F for Fake (1973) — a far more radical treatment of the same idea that forgery and art may not be separable categories.

Themes

Authenticity and its uses. The film's governing joke is that the fake Venus is beautiful, the fake Van Goghs give pleasure, and the only thing a certificate of authenticity secures is price. Value here is a social agreement, not a property of objects.

Inheritance as complicity. Nicole is not a thief by temperament; she is a daughter. The heist is filial duty, and the film's warmth toward Bonnet is inseparable from her inability to expose him.

Love as mutual con. Both leads are performing a false profession, and both are more legible to each other than to themselves. The romance resolves when the deceptions cancel out.

Appetite versus connoisseurship. Wallach's collector wants to own; the Bonnets want to make. The film has affection for making, including dishonest making, and comic contempt for acquisition.

Reception, canon & influence

Contemporary notices were broadly favorable on the stars, the photography, and the design, and broadly skeptical about length and pace — a pattern consistent across the film's reviewing history, though I will not attribute specific phrases here. It has never been treated as major Wyler: the standard biographical and critical literature on the director, including Jan Herman's A Talent for Trouble (1995) and Gabriel Miller's later study, handles it briefly as a late, light entry after the heavier achievements of The Best Years of Our Lives (1946), The Heiress (1949), and Ben-Hur. Its durable life has been on television and home video, where it functions as a comfort perennial and as a primary document of the Hepburn–Givenchy collaboration; the white lace dress and the black-and-white eyelet ensemble remain among the most reproduced images of 1960s screen fashion.

Backward. The film draws on Lubitsch's thief-lovers, on Rififi and Topkapi for the mechanics of the silent theft, on Charade for tone and personnel, and on the van Meegeren case for its premise. Trauner's design carries the memory of 1930s and 40s French studio filmmaking; Dalio's presence carries Renoir.

Forward. Its influence is diffuse rather than direct, but the template it perfected — the art heist as romantic comedy, security systems as flirtation, the caper in which nobody is hurt and the wrong party keeps the loot — runs through The Thomas Crown Affair in both its 1968 and 1999 forms, Entrapment (1999), Ocean's Twelve (2004), and the 2012 remake of Gambit. The alarm-defeat set piece belongs to a lineage that reaches Mission: Impossible (1996). And the film's philosophical joke — that the forgery is as good as the thing — was taken up far more seriously by Welles in F for Fake and, differently, by Abbas Kiarostami in Certified Copy (2010).

What the film ultimately preserves is a mode of studio filmmaking at the last moment it was possible: a great director, a great cameraman, a great designer, and two enormous stars, all applied with total professionalism to a story about how little authenticity actually matters.