What follows is an expanded version of a paper I delivered at the 2023 Society of Cinema and Media Studies Conference. I reproduce it here because this paper is really intended for anyone interested in writing critically about film as art rather than cultural artifact, however valid the latter view may be in its own sphere. Specifically, this essay presents a few key concepts developed in Gilles Deleuze’s Cinema 2: The Time-Image—concepts whose practical, film-critical use may sometimes be obscured by the rhetorical difficulties of Deleuze’s prose. This essay is an attempt to make them somewhat less obscure.
J. L. Austin once said that every philosophical claim consists of the part where you say it and the part where you take it back. I’m no philosopher, but I thought I’d try out the formula. And my claim goes something like this: “Gilles Deleuze’s Cinema 1: The Movement-Image and Cinema 2: The Time-Image constitute the most significant development of classical film theory since André Bazin.” Apart from integrating the work of numerous film theorists, such as Béla Balázs and Sergei Eisenstein, Deleuze’s Cinema books “complete” some of Bazin’s major theses. Most notably, Deleuze’s movement-image and time-image regimes map onto Bazin’s distinction between a “classical” and a “modern” cinema, while both broadening and better specifying that division.
I take the claim back, though, to note that on the whole, the books have not been received in this way. The Cinema books have generated many notable academic engagements, but these tend see the books either as footnotes to Deleuze’s more properly “philosophical” work, or in relation to Anti-Oedipus and A Thousand Plateaus. In the wider realm of film critical engagement, the Cinema books are not a recognizable point of reference in the way that Bazin’s What Is Cinema? continues to be. This is somewhat anecdotal, but in non-academic circles, I think one is still more likely to encounter Eisenstein’s “overtonal montage” or Tarkovsky’s “time-pressure” than terms like “movement-image” or “time-image,” still less their specific varieties.
So rather than exposit the Cinema books for those already interested in them, I will instead sketch out some of their organizing concepts, while trying to motivate their film-theoretical use with examples and illustrations. Specifically, I want to discuss what Deleuze in Cinema 2 calls “chronosigns.”
Time as Empirical Succession
To understand the function of these chronosigns, however, it’s first necessary to get a broad sense of Cinema 1. There, the big organizing principle is that in the movement-image regime, we have only an indirect image of time. In this regime, movement is what the cinema gives us directly, and it’s from movement that we infer or deduce the passage of time. Deleuze refers to the ancient philosophers’ definition of time as the number of movement. But in a way the reference is unnecessary, because this indirect view of time is the conventional one. It’s what we think of when we say chronological or clock time, whether that’s measured by the revolutions of the sun, the ticking of a mechanical watch, or the oscillations of a Cesium atom. Movies rarely measure time in so precisely quantitative a manner. But as in Dr. Mabuse, the Gambler (1922), where the movements of a briefcase and a stock market crash are linked by a criminal mastermind glancing at his pocket-watch, we see how time becomes intelligible only indirectly, and still depends on movement. Cinema 1 presents six different types of movement-images, the details of which I won’t go into here. The main thing to note is that no matter the type, we’re still dealing with an indirect image of time. In the movement-image regime of Cinema 1, time is viewed as an endless succession of nows—in terms of what Deleuze calls empirical succession.
At this point one might ask, well, what other view of time is there? If the movement-images of Cinema 1 give us an indirect image of time, then we should expect Cinema 2 to deal with direct images of time. And sure enough, Deleuze there introduces the seemingly odd notion of a “direct time-image.” We can easily understand what it means to see movement directly. But what does it mean to “see” time directly?
This is the story of Cinema 2, and the three chronosigns Deleuze introduces are the concepts that show us what it means to “see” time directly. The chronosigns “mark the presentation of the direct time-image.” In the time-image regime, we are concerned not with external empirical succession, but with the “internal relations of time.”
First Chronosign: Sheets of Past
Je t’aime, je t’aime (1968), Peppermint Candy (1999), (nostalgia) (1971)
Let’s consider the first chronosign: “sheets of past.” When we think of the past in movies, we’re typically led to the conventional flashback, which Deleuze defines, I think uncontroversially, as “a closed circuit which goes from the present to the past, then leads us back to the present.” Now if I were to ask whether All About Eve (1950) contains flashbacks, it is hard to imagine that anyone would answer no. If I then asked about Citizen Kane (1941), here again most everyone would agree, while perhaps adding that unlike All About Eve, Kane’s flashbacks don’t link up to any present-tense action—Kane is dead, and the mystery of Rosebud is revealed in a sequence that has no witness. If I then asked about Alain Resnais’s Je t’aime, je t’aime (1968), however, responses may start to divide. In that film, a time machine throws a suicidal man back into random moments of his past, which he is forced to relive fragments of; and it would seem odd, I think, to call what we see there flashbacks. With Resnais’s Last Year at Marienbad (1961) the issue is compounded. In that film, it’s hard to be sure how to even mark a present from which a flashback may be said to depart or return. There is no determinate “now” from which “last year” can be measured.
Empirical succession depends, by definition, on the fact that everything we see can be timestamped, that all occurrences are theoretically datable: However ambiguous or obscure a film may be, its events can be reconstructed into a linear, chronological timeline. In this view, past and future are simply former presents and presents to come. What a flashback shows us, then, is a “slice” of a former present, usually preceded by some device—a dissolve, for example—that signals that we are about to see a past moment. As the above films should illustrate, however, there are clear limitations to this chronological view, which may work well enough for All About Eve but makes nonsense of Marienbad. For Deleuze, the timeline model simply cannot explain the paradoxes and contradictions that arise when we confront the workings of memory.
It’s partly for this reason that Deleuze introduces the notion of the “past in general” or the “pure past,” which, following Henri Bergson’s Matter and Memory, he pictures as an inverted cone. The claim is that if one fixed a point on a film’s chronological timeline and called it the present, S, this infinitesimal moment coexists with the entirety of the past, pictured as a cone that extends upwards without limit. In this Bergsonian model, the past is not “behind” the present moment, as in a chronological timeline, but is, so to speak, “attached” to the “actual present,” S, which thus appears like an infinitely contracted past. Between this point S and the pre-existent past in general are so many “sheets” or “regions”: the AB circles in the diagram. These are “pure recollections” which “insist” on the present because they could be actualized in it but are not. These sheets of past are “virtual” not because they don’t exist, but because they have not been “actualized” in the present by being “brought down” from the cone, so to speak, in the form of recollection-images or flashbacks. Recollection-images, then, are pure recollections which are actualized and thereby brought forward into the chronological timeline: A person recalls something, actualizing a pure recollection into a recollection image, and then acts according to what they have remembered. The salient point is that conventional flashbacks, or recollection-images, are derivative of pure recollections. They depend, in short, on the first chronosign: sheets of past.
Now I won’t try to convince you of all this as a metaphysical truth, though Deleuze does insist that “we should have no more difficulty in admitting the virtual insistence of pure recollections in time than we do for the actual existence of non-perceived objects in space.” I’m mainly interested in the chronosign as a film-theoretical principle, and I think its explanatory power is best demonstrated by simply looking at different films. For instance, we can now say that Je t’aime, je t’aime does not in fact include any flashbacks. When the time-machine malfunctions, and the man, Claude, is forced to relive different fragments of his life, he is thereby cut off from empirical succession: The film effectively pins or freezes the point S, the “actual present,” on the chronological timeline. It then explores the pure past, showing as Claude relives different moments from his past, but each time from the perspective of a different “sheet” or “region” of the cone. Across the runtime, there’s a question of whether Claude did or did not kill his lover Catrine. And our ability to answer this question is, as Deleuze puts it, “probabilistic” because it’s a function of the pure past, whose internal relations change each time we move to a different sheet. Each time we are thrown into a different part of the cone, the relations between the sheets are continually redrawn, and our sense of the entirety of the past changes. The film gives us a direct time-image because in moving between these regions of the past, we “see” not external empirical succession, but the internal relations of time itself.
Another movie worth considering here is Lee Changdong’s Peppermint Candy (1999). The film starts with a man about to commit suicide by standing in front of a moving train. Right before the train hits him, the image freezes. The film then moves backwards, presenting seven discrete sections, each dated and titled, from his past: The first section is dated Spring 1999, from there we move three days into the past, then to Summer 1994, Spring 1987, and so on. Given the setup, we are naturally led to ask, “What led this man to kill himself?” And as we move from section to section, there’s always some object that carries over—a camera, or the peppermint candy of the title—forming a link between them. But the further away we get, the more clearly incapable this material chain becomes of explaining anything. At the same time, we notice that each new episode we see changes our total sense of the man’s actions, giving us a more holistic image of his personality and history.
Peppermint Candy thus permits two different ways of explaining his suicide. The film’s opening freeze-frame fixes the point S, the “actual present,” very clearly, which then allows us to see things in terms of either the chronological timeline or the pure past. The first perspective sees things in terms of empirical succession: Every section is a former present, and we simply order what we see in a material, causal chain of action. And viewed in this way, the film might seem rather deterministic, tendentious, and contrived. The second perspective, though, sees his actions as a function of his entire past: Each dated section gives shape to the whole of the past, and each time we are shown another episode, the total shape changes, because new relations form between the different sheets. From this perspective, his history, and therefore his actions, are seen in terms of the pure past—they are presented in a direct time-image.
As a final example, consider Hollis Frampton’s (nostalgia) (1971). In the film, we hear a voice (Michael Snow’s) narrating a story about a photo, while another photo is being burned on a hot plate in front of us. The catch is that we hear each story before we see the corresponding photo burned on the hot plate. If (nostalgia) strikes me as an especially elegant image of this chronosign, it’s because every time we try to engage with the photo in front of us, we find ourselves thrown back into the past—a past whose connection to the present is literally burning away and turning to ash.
Second Chronisgn: Peaks of Present
The Discreet Charm of the Bourgeoisie (1972), Certified Copy (2010), The Day He Arrives (2011)
With the chronosign we just dealt with, the pure past stands in for the whole of time. But this leads Deleuze to ask a different question: “Can the present in turn stand for the whole of time?” He answers yes—so long as we manage to separate the present “from its own actual quality, in the same way that we distinguish the past from the recollection-image which actualized it.” Just as we contrasted “recollection-images” with “pure recollections,” we can see the present detached or cut off from actualization in terms of empirical succession. Rather than explore a succession of events unfolding in time, we could explore all the aspects or accents of a single present event. This would give us the second chronosign, “de-actualized peaks of present.” For inspiration Deleuze turns to St. Augustine’s notion of the “present of the past, present of the present, present of the future.” But as is often the case, I find it easier to understand the concept through examples.
Deleuze’s clearest one comes from Buñuel, particularly in the contrast between The Exterminating Angel (1962) and The Discreet Charm of the Bourgeoisie (1972). In The Exterminating Angel, a group of wealthy Spaniards sit down for a dinner party, but later find themselves unable to leave because of some mysterious barrier. After roughly two days, when the dinner guests are finally able to leave, they go to a church and pray, and it’s at this point that a revolution gets underway. We hear the sounds of gunfire, and the guests are probably killed. In Discreet Charm, we have a similar set-up. A group of bourgeois French friends keep intending to have a dinner party, only to have the meal be eternally deferred. As in Exterminating Angel, the end of Discreet Charm sees them mowed down by gunfire.
Both films are concerned with class conflict and make clear use of structural repetition. The crucial difference is that in The Exterminating Angel, we are still in the realm of empirical succession: From the present event, we infer a kind of cycle of history stretching out into the future, marked in this case by a revolution breaking out and the collapse of the existing social order. This is an indirect image of time, and so it makes sense that Deleuze discusses the film in Cinema 1. In Discreet Charm, by contrast, Deleuze argues that we don’t have succession but simultaneity. As he writes in Cinema 2, Discreet Charm “shows less a cycle of interrupted meals than different versions of the same meal in irreducible modes and worlds.” Rather than a succession of different events unfolding in time, as in The Exterminating Angel, Discreet Charm shows us different aspects of the same event, implicated with each other, as if superimposed on top of each other simultaneously. These are the accents or peaks of present.
To take another example, consider Abbas Kiarostami’s Certified Copy (2010). Set in Tuscany, the film opens with an academic (William Shimmel) giving a lecture on his new book, which is attended by an antique dealer (Juliette Binoche), who arrives late (and rather disruptively) with her son. Shimmel and Binoche’s characters spend an afternoon together—and as we watch them, they seem at some points to be strangers who have just met, at others to be partners in a long-term marriage, and almost every variation in-between. The originality of Certified Copy is that it doesn’t resolve these possibilities, but allows them to coexist simultaneously, all at once, within this one afternoon. If the question of this chronosign is “Can the present stand in for the whole of time?”, the analogous question in Certified Copy is “Can an afternoon stand in for the entirety of a marriage?” The film does not present us a dramatic progression of two people developing a relationship in time, as we would get in the movement-image regime. Rather, it presents us with different accents of a relationship, from the halting awkwardness of a strangers’ meeting to the familiarity of a long-term marriage, all contained in this one afternoon encounter.
As a final example, we may look at Hong Sangsoo’s The Day He Arrives (2011). In the film, a former director arrives in Seoul to meet a friend—but when the friend doesn’t show up, he ends up staying a few more days. What is striking about the film is that each day plays out much like the last, and so we get an uncanny sense of déjà vu. The director’s arrival seems to cycle and repeat again and again, with every day feeling like the day he arrives. The film as a whole has a rather downbeat trajectory of failure and disappointment, but as Dan Sallitt writes, the impression the film leaves us with is “of the same loss playing out at different speeds and with different circumstances.”
Third Chronisgn: Time as Series
Right Now, Wrong Then (2015), Yourself and Yours (2016), Chronicle of a Summer (1961), Standard Gauge (1984)
Thus far we have dealt with two chronosigns: sheets of past and peaks of present. These concern what Deleuze calls the “order of time, that is, the coexistence of relations or the simultaneity of the elements internal to time.” Each chronosign shows how we can see past and present as “virtual” and “de-actualized,” detached from empirical succession, and therefore in terms of a direct time-image: The films we have considered specifically negate or diminish our sense of forward temporal movement. But what of the future? How can one deal with the passage of time without recourse to external empirical succession?
Deleuze’s answer to these questions involves the third chronosign: time as series. This he relates to how “time has always put the notion of truth into crisis,” as in the ancient paradox of future contingents, the textbook example of which involves a naval battle which may or may not take place the next day. As always, though, I prefer looking at film examples—in this case to Hong’s exemplary Right Now, Wrong Then (2015). In the film, an arthouse director (Jung Jaeyoung) has a chance encounter with a budding artist (Kim Minhee), the story gimmick being that the scenario plays out twice, in two discrete parts. At the end of the first (titled “Right Then, Wrong Now”), they don’t get together; at the end of the second (titled “Right Now, Wrong Then”), they do. The film’s dual titles are typically read as describing a lateral relation across the two sections. But when considering the so-called paradox of contingent futures, they are more instructively seen in relation to possibility and outcome—the before and after of each segment, as it were. At the start of each section, it is true that the director and the artist may end up together. But when they separate at the end of the first part, the proposition that was “right then” gives way to an outcome that is “wrong now.” In the second part, when the couple gets together at the close, the outcome that is “right now,” makes the original statement “wrong then,” because it’s no longer possible that they may not get together.
Now it’s easy to dismiss this paradox as idle sophistry. What’s more difficult to ignore is the challenge of reconciling truth with a form of time. Delezue’s complex, highly engaging account of this difficulty runs through Leibniz’s incompossible worlds, Borges’s forking paths, and Nietzsche’s will-to-power—but his main point, as I understand it, is the discovery of an essentially falsifying power in time. Every surge forward in time, every fork in the path, creates a paradoxical relationship between past and present, before and after. No longer extricable from the passage of time, truth becomes a function of how one marks out befores and afters. Truth becomes a function of how one constitutes a series.
In Right Now, Wrong Then, Hong goes about as far as one can to keep the two parts in separate worlds. Accordingly, the film may rightly be read as presenting us two simultaneous peaks of present. But in another sense, the film’s symmetric, neatly demarcated structure only clarifies the impossibility of escaping the consequences of the third chronosign—for at the end of Right Now, Wrong Then, the issue of ordering alternatives remains. As I’ve written previously, it’s impossible for Hong to avoid affirming one outcome over the other, making the film comic rather than tragic, optimistic rather than pessimistic. Considered as a series, the film is therefore distinct mainly for how clearly it marks out its terms. But in the end, there is no reason a film’s series must be marked out in any explicit way, nor that it need even show its forking paths.
Yourself and Yours (2016) multiplies and compounds the consequences of this chronosign. Near the film’s start, a painter Youngsoo (Kim Joohyuk) accuses his girlfriend Minjung (Lee Yooyoung) of getting drunk in public yet again, despite having promised to cut down on her drinking, while she in turn maintains her innocence. After the fight, they decide not to see each other for a while, and during their time apart, Younghoo sullenly pines for Minjung and reflects on his commitment to her. “Minjung” for her part goes out with other men—scare quotes because in each of these encounters, in which the man initially claims to recognize her, she in turn claims not to be the person they think she is, and at one point tells a man that she is Minjung’s twin. It’s something like a joke about object permanence at the expense of a possessive male ego: Each time Minjung goes out of the man’s immediate vicinity, she literally becomes a different person. Accordingly, each encounter involves a mix of possibilities: brazen lies or alcohol-induced memory lapses on Minjung’s part, lame flirtation devices or genuine cases of mistaken identity on the man’s, and the outside chance of a twin sister roaming the city.
One could of course see the entire film as Minjung piling lie on top of outrageous lie, doubling down with each new encounter—which, as Dan Sallitt observes, is “nearly the formula for a silent comedy of escalation, in which the game would be to elaborate and re-contextualize a gag until the narrative threatens to collapse under its weight.” This account requires, though, that we be able to resolve the dynamics of each encounter, and then slot them into a coherent timeline—and Yourself and Yours is far stranger and more ambiguous than that. Since each new encounter multiplies the narrative’s forking paths, any attempt to resolve a reading of one scene changes the very possibilities that are at play within the others. The consequences of fixing one term in the series ripple backwards and forwards in time, effectively changing the possibilities of all the rest. It is a measure of the film’s intricate structure that it is impossible to recount its narrative without ruling out a set of other possible accounts. (That Hong attributes two early dream sequences to Youngsoo further compounds these difficulties.) The beauty of the film’s ending is that when Youngsoo finally encounters or re-encounters “Minjung,” he accepts her completely—every possible version of her. Their final scene together conveys a vision of trust and change that may recall classic remarriage comedies such as The Awful Truth (1937) and The Lady Eve (1941). But however great those films—and they are great—they all conform to an indirect image of time as empirical succession. Hong’s achievement with Yourself and Yours is to present a remarriage comedy in a direct time-image.
Yourself and Yours shows very clearly how character transformation is no longer explicable in terms of external empirical succession. Under the conditions of the direct time-image, the characters themselves can take hold of the falsifying power of time. As in Jean Rouch and Edgar Morin’s cinéma-vérité classic Chronicle of a Summer (1961), it is the characters’ “becomings,” as Deleuze puts it, that now mark the befores and afters of a series. The film begins with the filmmakers asking various people on the street a simple question: “Are you happy?” Very quickly, though, things start to multiply. Scenes continually give way to reflections and analyses of previous ones. The subjects of a given sequence will later reflect on the very experience of filming or on the resultant footage, in effect falsifying their previous positions, and even their previous selves. Chronicle’s famous final scenes show the subjects and filmmakers reflecting on, dissecting, and re-describing the film itself. And it is these repeated acts of re-description that prevent the film from being unified in terms of empirical succession, creating in each scene an incommensurability between the “before” and “after”—a gap that can only be bridged by a character’s becoming.
As these three examples should make clear, there is no one way by which the terms in a series should be constituted. Morgan Fisher’s Standard Gauge (1984) goes even further, showing that one need not even cut or move the camera. In one fixed, continuous shot, Fisher exhibits scraps of discarded film collected from his time in Hollywood’s post-production sector in the 1970s and 1980s, using their personal import to motivate an account of the forking paths that constitute his life. Even more significantly, he speaks also to the technical and historical significance of the scraps—mainly in relation to the standardization of 35mm in the early twentieth century, and the auxiliary Hollywood production practices that emerged as a result, such as the insertion of so-called “China girl” frames used to standardize colour. In this way, Fisher allows the notion of time as series to take on a reflexive, literally material dimension. Looking to the scraps and detritus of the Hollywood system—to those terms in the series that were never meant to be seen by a paying audience—he draws out possibilities discarded and ruled out by the very constitution of the standardized series. It is part and parcel of Fisher’s canny construction that Standard Gauge was shot with a 1000-foot reel of 16mm film, and therefore runs exactly thirty-five minutes, a duration far exceeding the durational capacities of 35mm. Standard Gauge is in effect an alternative history: a series whose very constitution falsifies the dominance of the standard gauge.
The film’s very existence points to an industry whose rapid changes would only hasten the obsolescence of 16mm. And though Fisher never directly broaches the subject, he seems to allude to it when he uses a scrap of Edgar G. Ulmer’s The Naked Dawn (1955) as an excuse to discuss a scene from another Ulmer film, the noir classic Detour (1945), in which an escaped fugitive predicts his eventual arrest: “His shifting to the future tense makes the scene jump forward, revealing as a certainty the outcome that the character’s fatalism has long foreshadowed. And yet this disjunction in time, a kind of ellipsis into the future, takes place within a single continuous shot.” It is characteristic of Fisher’s “literalist” tendencies that these words describe not just the composition of Standard Gauge, but his understanding of time as series. For in the very act of internalizing his obsolescence in Standard Gauge, Fisher, like Detour’s fatalistic criminal on the lam, has already gone beyond the coordinates of the classical regime. He sees the notion of before and after in terms of the third chronosign, time as series—no longer as “a matter of external empirical succession,” but of “the intrinsic quality of that which becomes in time.”
The Pure Good of Theory
Was it that—a sense and beyond intelligence?Could the future rest on a sense and be beyondIntelligence? On what does the present rest?
—Wallace Stevens, The Pure Good of Theory
It’s at this point worth clarifying that the three chronosigns of Cinema 2 are less discrete categories than possibilities of exploration. It should be possible to see—in fact we should expect to see—filmmakers explore not just one chronosign, but configurations of two or three. Deleuze’s treatment of Last Year at Marienbad shows how the film develops a tension between the tendencies of Rensais, whom he associates with sheets of past, and those of co-writer Alain Robbe-Grillet, whom he associates with peaks of present. And Hong’s oeuvre constitutes arguably the most accomplished exploration of the interpenetration of the three chronosigns in contemporary cinema—an achievement that risks being undervalued by the continual focus on the autobiographical aspects of his work, as if each new film were just another drama of artistic crisis. It would be as if writing on Borges focused only on the fact that his stories routinely feature manuscripts and books.
Hong’s example should also demonstrate the limitations of thinking about time in the cinema purely in terms of empirical succession, and hence the practical usefulness of the chronosigns as film-critical terms. It is one thing to give a negative account of what a film does, and say that it “violates,” “departs from,” or otherwise negates chronological time. But it is quite another to give a positive account of what relations of time the film shows us. The latter is what the chronosigns of Cinema 2 permit us to do.
Discussions of time that depart from empirical succession or clock-time are too often (and I think too quickly) derided as idle “metaphysical” speculation. But one need not accept Deleuze’s metaphysics to find value in the film-theoretical use of his chronosigns, and to see how useful they are in discussing the relations between movement and time in the cinema. And perhaps apart from some of the rhetorical challenges posed by the Cinema books, I see no reason why the chronosigns cannot enter the film-critical lexicon in the manner of the distinctions drawn by Bazin and other classical film theorists. In any case, the conviction that underlies this aspect of Deleuze’s work, as I understand it, is “transcendental” in the sense that Kant originally gave the term—the sense, as he put it in the Critique of Pure Reason, that we can speak of space and time “only from the human standpoint.”