David Bordwell's website on cinema   click for CV




Perplexing Plots: Popular Storytelling and the Poetics of Murder

On the History of Film Style pdf online

Reinventing Hollywood: How 1940s Filmmakers Changed Movie Storytelling

Film Art: An Introduction

Christopher Nolan: A Labyrinth of Linkages pdf online

Pandora’s Digital Box: Films, Files, and the Future of Movies pdf online

Planet Hong Kong, second edition pdf online

The Way Hollywood Tells It pdf online

Poetics of Cinema pdf online

Figures Traced In Light

Ozu and the Poetics of Cinema pdf online

Exporting Entertainment: America in the World Film Market 1907–1934 pdf online


Hou Hsiao-hsien: A new video lecture!

CinemaScope: The Modern Miracle You See Without Glasses

How Motion Pictures Became the Movies

Constructive editing in Pickpocket: A video essay


Rex Stout: Logomachizing

Lessons with Bazin: Six Paths to a Poetics

A Celestial Cinémathèque? or, Film Archives and Me: A Semi-Personal History

Shklovsky and His “Monument to a Scientific Error”

Murder Culture: Adventures in 1940s Suspense

The Viewer’s Share: Models of Mind in Explaining Film

Common Sense + Film Theory = Common-Sense Film Theory?

Mad Detective: Doubling Down

The Classical Hollywood Cinema Twenty-Five Years Along

Nordisk and the Tableau Aesthetic

William Cameron Menzies: One Forceful, Impressive Idea

Another Shaw Production: Anamorphic Adventures in Hong Kong

Paolo Gioli’s Vertical Cinema

(Re)Discovering Charles Dekeukeleire

Doing Film History

The Hook: Scene Transitions in Classical Cinema

Anatomy of the Action Picture

Hearing Voices

Preface, Croatian edition, On the History of Film Style

Slavoj Žižek: Say Anything

Film and the Historical Return

Studying Cinema


Book Reports

Observations on film art

What happens between shots happens between your ears

Monday | February 4, 2008   open printable version open printable version

DB here:

In Number, Please? (1920) Harold Lloyd plays a lovesick boy who’s been jilted by his girl. Moping at an amusement park, he sees her arrive with a new beau.

num-pl-test-1-225.jpg num-pl-test-2-225.jpg

He shifts to another spot to watch them. When she notices him, she scorns him, and he reacts.

num-pl-test-4-225.jpg num-please-test-5-225.jpg

She and the escort stroll past, then she turns and cuddles up to him, making sure Harold notices.

cap024-225.jpg cap025-225.jpg

Her flirting precipitates the rest of the action in this very funny short.

In this scene from Number, Please? Harold and the couple aren’t shown in the same frame. The action is built entirely out of singles of Harold and two-shots of the couple, with an especially emphasized close-up of the girl’s snooty reaction. The sequence is rapidly paced and carefully matched in angle; note the shift in eyelines as the girl and her beau walk a little way and then she looks back at Harold.

This approach to building a scene was consolidated in American studio cinema in the late 1910s, as we noted recently, and it soon spread around the world. One of the places it caught on most fervently was Soviet Russia.


Kuleshov glories in the gaps

The great director and theorist Lev Kuleshov always claimed that he and his associates learned the power of editing from American cinema. Russian films were slowly paced, consisting of long tableaus occasionally broken by an inserted closer view of an actor. (For examples, see my Bauer blog entry from the summer.) Kuleshov noted that the Hollywood films brought into Russia grabbed audiences’ interest more intently, and Kuleshov attributed this effect to the fact that the Americans exploited editing more fully, creating the scene out of many shots.

Kuleshov’s example was the formulaic scene of a man sitting at his desk and deciding to commit suicide. The Russians, Kuleshov claimed, would handle this all in one distant framing, with the result that the key actions were just part of the overall view. By contrast, Americans would shoot the scene in a series of close-ups: the man’s face, his hand taking a pistol out of a desk drawer, his finger tightening on the trigger, and so on. This gave the scene a powerful concreteness, and was cheaper to film besides (no need to have a full set).

But most American filmmakers didn’t create the scene wholly out of close-ups. Typically there would be an establishing shot before the action was broken down into detail shots. The process has come to be called analytical editing. (We discuss it in Chapter 6 of Film Art.) As the label implies, the cutting analyzes an orienting view into its important details.

Less commonly, as in our Number, Please? example, American directors could create a scene entirely out of separate areas of space, without ever showing a master framing. This technique, usually called constructive editing, remains common today as well, especially in action scenes.

While praising analytical editing, Kuleshov was particularly taken with constructive editing, because that shows that cinema can call on the spectator’s tacit understanding to assemble the separate shots. Kuleshov realized that we will build a sense of the scene’s space and action out of separate shots without need for the comprehensive view supplied by an establishing shot.

What the Americans developed, the Soviets thought seriously about. Around 1921, Kuleshov and his students mounted some experiments, several of which he discusses in his books and essays. He probably didn’t need to experiment; the American films were full of examples. Indeed, the Number Please? passage is more intricate than any experiment Kuleshov seems to have mounted. But he had a bit of the engineer about him, and he sought to break the technique into its simplest components.

For one thing, constructive editing offered production efficiencies. You could film two actors separately, at different times, and then cut them together. Further, Kuleshov saw that editing could abolish real-world constraints. It created events that existed only on the screen, with an assist from the viewer’s mind.

This is perhaps best seen in his experiment involving an “artificial person.” Evidently it wasn’t a case of constructive editing, because it seems to have begun with an establishing shot. The first shot shows a girl sitting at her vanity table putting on makeup and slippers. A series of close-ups of lips, hands, legs, and the like were derived from different women, and they were edited together to create the impression of a single woman. (Something of this effect survives in the idea of a body double in contemporary films and TV commercials.) But the point is that the woman on screen, made out of different parts, doesn’t exist in the real world.

The same possibility could apply to geography, if we delete the establishing shot. As Kuleshov describes his experiment, we initially get a shot of a woman walking along a Moscow street. She stops and waves, looking offscreen. Cut to a man on a street that is in actuality two miles away. He smiles at her and they meet in yet a third location, shaking hands. Then together they look offscreen; cut to the Capitol in Washington DC. Kuleshov saw the potential for imaginary geography as both a useful production procedure and a demonstration that editing could create a purely cinematic space, one not beholden to reality.

Kuleshov’s most famous experiment, the one he identified with the “Kuleshov effect” proper, involves a stock shot of the actor Ivan Mosjoukin, taken from an existing film. In his writing he’s rather vague and laconic about the results.

I alternated the same shot of Mosjoukin with various other shots (a plate of soup, a girl, a child’s coffin), and these shots acquired a different meaning. The discovery stunned me—so convinced was I of the enormous power of montage. (1)

Kuleshov’s pupil the great director V. I. Pudovkin offers a different description of the shots: a plate of soup, a dead woman in a coffin, a little girl playing with a teddy bear. He also goes farther in reporting how the audience responded. People read emotions into the neutral expression on Mosjoukin’s face.

The audience raved about the actor’s refined acting. They pointed out his weighted pensiveness over the forgotten soup. They were touched by the profound sorrow in his eyes as he looked upon the dead woman, and admired the light, happy smile with which he feasted his eyes upon the girl at play. But we knew that in all three cases the face was exactly the same. (2)

Now it isn’t only geography or a human body that has been created by editing; it’s an emotion.

These experiments have been poorly documented, and only a couple have survived. One consists of fragments of the imaginary geography exercise. Here is Alexandra Khokhlova waving, but we don’t have the answering shot of the male actor responding. The two meet at the foot of a statue.

k-effect-khokhlova-waves-200.jpg k-effect-master-200.jpg

After the two shake hands, they look up and out of the frame, but unfortunately we lack the shot of the Capitol.

k-effect-handshake.jpg k-effect-2-shot-200.jpg

Kristin and other scholars have written more about these surviving fragments, and their essays are published along with Kuleshov’s proposal for funding the experiments and his wife Khokhlova’s memoir of filming them. (3)

It’s worth taking these prototypes of constructive editing apart a little bit. Clearly, there are several cues that prompt us to see the shots as continuous.

One cue is a common background, or at least a consistent one. Kuleshov thought that sometimes a neutral black background worked best, especially for close shots, as you can see with the handshake shot. Another cue is roughly consistent lighting from shot to shot. Yet another is the presumption of temporal continuity; no moments seem to be omitted in the cut from shot A to shot B. It never occurs to us to consider that Kuleshov’s man is looking at something hours or days before the soup is laid on the table in the second shot.

One of the most important cues goes unmentioned by Kuleshov: the very act of looking. Like most commentators, he seems to take it for granted. Yet it’s crucial in prompting us to imagine an overall space in which the actions take place. Knowing the real world as we do, we can infer that if you’re close enough to watch someone, both people are probably in a shared space, such as the arcade strip in the amusement park in Number, Please?

Another cue is facial expression. In his soup/girl/coffin sequence, Kuleshov supposedly picked a shot of Mosjoukin that doesn’t have a clearly identifiable expression. If Mosjoukin was smiling in his interpolated shot, he would presumably seem not grieving but wicked. Normally, though, performers seen in the single shot are expected to express the appropriate emotion more fully, in order to specify what we take the characters’ mental states to be. Our sequence from Number, Please? makes sure we understand the drama by giving the actors unambiguous expressions.

Finally, in the Kuleshov prototypes each shot should be simple. Its action can be stated in a brief sentence. A woman waves. A man responds. A man looks. A plate of soup sits on a table. Even in Number, Please? we can summarize each shot: Harold looks off. His former girlfriend disdains him. That’s to say that the shots are composed to present only one, quickly grasped point of interest.

Filmmakers don’t need to tease apart all these cues; they use them intuitively. Soon after Number, Please?, Hollywood filmmakers were creating amazing passages of eyeline-match editing—the most virtuoso being probably the racetrack sequence in Lubitsch’s Lady Windermere’s Fan (1926). Within a few years of Kuleshov’s experiments, Soviet filmmakers were creating their own masterworks, like Battleship Potemkin (1925) and Kuleshov’s By the Law (1926). Benefiting from a very compressed learning curve, filmmakers took constructive editing to new heights.


Constructive editing, dissolved relationships

Sometimes constructive editing is used to save a scene when production goes astray. When Doug Liman reshot the climax of The Bourne Identity, Julia Stiles couldn’t be on set, so singles of her taken from the first version were wedged in among the retakes, to create the impression that she was watching Jason confront his controller. More positively, a carefully calibrated constructive editing is central to a sequence I analyzed a while back in In the City of Sylvia. For over 100 shots, the spatial relations are built up without an overall establishing shot. (4)

Constructive editing can be used systematically throughout a film. A good example is Alan J. Pakula’s thriller Presumed Innocent (1990). Spoilers coming up!

Harrison Ford plays Rusty Sabitch, a prosecutor in the District Attorney’s office who becomes infatuated with a new woman on the staff. He has a brief affair with her, but after she’s broken it off she’s found brutally murdered. He has to investigate the crime without involving himself, but eventually he becomes the prime suspect.

At the start of the film before Rusty learns of the murder, Rusty and his wife Barbara are shown at breakfast, and establishing shots bring them together.


At the office Rusty learns of Carolyn’s death, and after he comes home, the conversation between Rusty and Barbara is treated without an establishing shot. Barbara knows about the past affair, and Rusty is wracked by guilt and shame. The cutting seems to reflect the fact that Carolyn’s death has revived the pain in their marriage.

cap024-325.jpg cap025-325.jpg

In a series of flashbacks, Rusty relives his affair with Carolyn. Pakula treats their early encounters by means of constructive editing. The common-background cue is especially helpful in this scene in a kindergarten.

cap038-325.jpg cap037-325.jpg

Only after Carolyn and Rusty cooperate and win a child-abuse case does the cutting’s rationale become clear. Pakula has saved the traditional two-shot framing for their moment of passion, as they make furious love in the office.


But soon their affair wanes and Rusty is reduced to watching Carolyn from across the street in point-of-view shots.

cap049-325.jpg cap050-325.jpg

After the flashbacks end, Rusty is investigated and eventually charged with Carolyn’s murder. In these scenes Pakula often situates Rusty and Barbara in the same frame, as if the threat to him has healed the breach between them.

cap054-325.jpg cap060-325.jpg

Yet in the film’s climax—and here is the big spoiler—they are pulled apart again. The last scene is a sustained monologue by Barbara. As Rusty listens, across twenty-six shots the two are never shown in an establishing shot.

cap063-325.jpg cap064-325.jpg

Contrary to the standard Hollywood ending, in which the loving couple embrace in reunion, here they are left divided.

Pakula’s use of constructive editing has effectively traced two arcs: the growth and dissolution of Rusty’s affair, and the reuniting and dissolution of his marriage. In such ways, what might seem a purely local effect, confined to a handful of shots, can create stylistic patterning across a film. The judicious use of constructive editing matches the dramatic development.


Godard of the gaps

Robert Bresson has made more varied and complex uses of constructive editing across a film, as I tried to show in Narration in the Fiction Film and in an Artforum essay. (5) A more radical approach, somewhat in the purely Kuleshovian spirit, can be seen in Godard’s films since the late 1970s. In presenting a scene, Godard often omits an establishing shot, so constructive editing takes over. But he makes the technique quite abrasive and ambivalent.

In watching films like Number, Please? and Presumed Innocent, we fill in the gaps between shots with ease. Godard, however, makes his images and sounds more fragmentary by equivocating about the Kuleshovian cues. The background elements and lighting don’t match entirely; time seems to be skipped over; glances and facial expressions are ambivalent. Adding to these factors, lines of dialogue slip in from offscreen. Godard does present a dramatic scene taking place, but he also creates a sense that images and sounds have been pried loose from their place in an ongoing action, floating somewhat free and functioning as objects of contemplation for their own sake. The cues that Lloyd insists on and that Kuleshov plays with are ones that Godard suppresses or makes ambiguous.

I’ve mentioned this tendency in a recent entry, but to elaborate a little, consider the science lecture in Hail Mary (Je vous salue Marie, 1985). A professor who has emigrated from an Eastern bloc country is explaining his theory that life on earth began and evolved because it was directed by extraterrestrials. No establishing shot of the classroom shows us where he, Eva, and two male students are located, so we have to construct a rough sense of their positions on the basis of a few cues. As Eva, perched on a windowsill, toys with a Rubik’s cube, we hear the professor’s lecture begin to speculate on whether life could have evolved spontaneously. His remarks about sunlight coincide with a burst of sun on her face.

After the Biblical title, “In those days,” we get a series of shots that allow us to apply our mental schema of a classroom lecture: attentive students looking off, a professor at the blackboard.


Lacking an establishing shot, we can’t specify how many people are in the class, nor indeed where Eva and her classmate are sitting—since the professor looks off in several directions during his talk.

At the end of his talk he remarks, still scanning the room, that we can presume that life exists in space. “We come from there.” At that point Godard cuts to the head of another student, seen from behind. The sproingy haircut is a little explosion of yellow, and it’s accompanied by a burst of choral music. And as the shot goes on, we start to notice that the professor is pacing in the background, out of focus.

The student, whom we’ll learn is named Pascal, asks a question (at least the slight head movement suggests that it comes from him), and the professor replies. Pascal scratches his head as the professor continues, still out of focus. If I had to choose one shot that condenses Godard’s strategy of suppression in this sequence, this shot would be my candidate.

At the end of the shot, the professor asks Eva to stand behind Pascal. Cut 180 degrees and somewhat farther back to Pascal, now seen from the front. The professor’s hand brings the Rubik’s cube into the shot and Eva comes up behind Pascal as the professor passes.


Later she and the prof will become lovers, but Godard lets them meet first as simply two torsos intersecting behind Pascal. The purpose is a demonstration that a “blind” agent can be steered toward a goal through simple yes/no commands. This models the professor’s theory that an alien intelligence could have guided evolution.

Pascal will twist the facets of the cube under Eva’s hints. Godard makes this a tactile, even erotic exchange, with the close-up of her by his ear and Eva saying, “Yes,” more urgently as Pascal’s hands arrive at a solution, in the close-up surmounting this blog entry.

The next two shots of the sequence center on the prof, who has exited frame right from the “three-shot.” Now he’s at the window, recalling the initial shot of Eva; but unlike her he’s little more than a silhouette. As crashing organ music is heard, he seems to be watching the experiment from afar. The second shot, an axial cut-in, coincides with the offscreen voice of a male student: “Were you exiled for these ideas?”


“These ideas, and others,” the prof replies. He says he’ll see the class on Monday, evoking the idea that he’s dismissing the offscreen students, and he turns his head slightly, though we can’t be sure they’re on his right. This shot will be held for some time as students quiz him more about his theory, and Eva asks him if he’d like to come over for a drink some evening. But we don’t see her say it. Godard cuts to a shot of Eva at the window, bathed in sunlight, opening and shutting her eyes as she slightly lifts and lowers her head.

As we see her, we hear the rustling of people leaving (so the class was evidently larger than three). And we hear him reply to her invitation, “That’s another story [scénario].” Are Eva and the prof looking at each other? We’re inclined to say yes, but her closed eyes and tilting head suggests someone lost in contemplation rather than engaged in conversation. Here the classic facial-expression cue is made indeterminate. We have no way of knowing if the prof’s line comes from offscreen, or is displaced from another point in time; maybe he has left the room. Such displaced diegetic sound occurs elsewhere in the film.

We don’t have to decide; our indecision is the point. By pruning away the most reliable cues, Godard wins both ways. We’re kept somewhat oriented to an intelligible action: a prof sets forth his far-fetched theory of human origins and a woman in his class invites him for coffee. This side of the balance allows us to feel that a story, however sketchy, is moving forward.

But the moment-by-moment texture of the scene allows the individual shots, gestures, and sounds to drift somewhat free. Each image takes on a more intrinsic weight, and the juxtaposition of picture and sound acquires a resonance that we usually call poetic. A shot of Eva in the sun playing with the Rubik’s cube, unanchored in time (during class? before class started?), invites us to apply metaphors, especially once we learn her name. Pascal’s thorny hair suggests not only extraterrestrials but the explosion of a nova. The silhouetted prof, detached from the mechanism he has set in motion, hints at an unknown deity watching the game play out according to his rules. Why do Godard films spawn long essays built out of erudite associations? Because the narrative progression relaxes and we can weave our own connotations out of what we see and hear.

If you don’t want to go down the expanding-association route, there’s another one open. When individual moments no longer accumulate ordinary dramatic pressure, we can savor the fugitive pleasures of the separate shots (light on face, lips by ear) and the patterns they form: flipover cuts, yellow hair and yellow facets, bookended shots of Eva at the window.

Those patterns, it should be clear, depend on our sensing a bump at every shot change, looking for a way to skip across the gap that Godard creates. The same belief that meaning and effect are born of gaps impelled Kuleshov too, and perhaps even Lloyd. If we pay attention to those gaps we can feel minds—both the filmmaker’s and ours—at work in them.

(1) Lev Kuleshov, “’50’: In Maloi Gznezdinokovsky Lane,” Kuleshov on Film, trans. and ed, Ronald Levaco (Berkeley: University of California Press, 1974), 200.

(2) V. I. Pudovkin, “The Naturschchik instead of the Actor,” Selected Essays, ed. and trans. Richard Taylor (Oxford: Seagull, 2006), 160.

(3) See Kristin Thompson, Yuri Tsivian, and Ekaterina Khokhlova, “The Rediscovery of a Kuleshov Experiment: A Dossier,” Film History 8, 3 (1996), 357-367.

(4) The sequence does begin with a long shot of the café, but it is so distant, crowded, and brief that it can’t really be said to establish the spatial relationships among the several characters we see.

(5) “Sounds of Silence,” Artforum International (April 2000): 123.

Comments are closed.

David Bordwell
top of page

have comments about the state of this website? go here