(Translation of the original essay from 1975. Translation published in "AFTERIMAGE", London, 1981. Meanwhile this essay has been expanded and its vocabulary partially changed. In the changed form it became integrated into Wyborny's "Elementary Editing Theory". The main difference in vocabulary is, that Wyborny refuses to use the terminology "grammar of the narrative film" any longer and substitutes it systematicly by the more neutral expression "narrative system". Also the systematic diagram from the end is strongly expanded.)


The following pages make up the first part of a work entitled 'Elementary Editing Theory: An introduction to the Grammar of the Conventional Motion Picture'. Within the framework of that essay, the contents of these pages expound certain premises, which serve as a basis for a more thorough-going analysis.

A) The Reality of the Steady Stare

The cinema's image of human beings: characters who stare into each other's eyes at length; characters who observe objects, landscapes and other characters with unmoving eyes; characters who steadfastly return every stare - if not, they are supposed to have something to hide. The thinker stares in the course of his thinking at a hypothetical point in the room without moving his eyes, and it is this which demonstrates his mental activity.

On the other hand, reality: characters who only accidentally look into each other's eyes: characters whose irises continually oscillate when they look at something carefully; characters who would not think of returning a stare; and a mental activity so intimately associated with uncontrollable eye-movement that you might surmise that it is this very eye-movement, continually wiping the reservoirs and cells free of the residues of previous mental efforts, that makes a fresh train of thought possible.

When interviewed, people move their eyes with nimble dexterity as they answer. If I am conversing with someone who does not move his eyes, I am sure that he is not listening to me. When I recite a poem, I can keep my eyes fixed on a single point for as long as I can perfectly recall the text. If I have to consider how to continue, I am obliged to move my eyes. It is no accident that actors move their eyes little in films. They have already learnt the words they are speaking by heart and should they move their eyes, the motion will be somewhat mechanical focussing on surrounding objects, on flowers, ash-trays, the edges of a desk.

The eye movement of a person who is thinking, on the other hand, does not focus on surrounding objects. It is unfocussed If I am thinking, I cannot simultaneously see something in sharp focus. If I am staring at a fixed point in space, even my field of vision disappears. To be able to see, I need movement. If this movement is not provided by the objects around me I must produce it through my iris. Conversely, people who move their eyes rapidly in a film are those who have something to conceal - criminals, neurotics and the insane. Narrative cinema transforms mental activity into an attribute of sub-criminal anomaly.

Narrative cinema treats interaction between characters as the visible interaction of rigid bodies with a well-defined momentum, a well-defined line of movement, and a well-defined shape. Shifts in momentum, line of movement and shape have a well-defined cause in a well-defined interaction with another well-defined body.

The main function of narrative grammar is to produce a natural connection between units of space and time not naturally connected. The yardstick for the naturalness of such a grammatical bond is the emergence of a space-time structure which is not internally contradictory at first sight. This is to say that implicit in every element of the emerging space-time structure is the virtual imprint of a geographical and a temporal co-ordinate, and these co-ordinates must not be contradictory. It would be contradictory, for example, for a rigid body to occupy two different geographic co-ordinates simultaneously.

The space-time constructions of narrative grammar are accomplished almost exclusively by a single operational principle, by the exchange of carriers of motion in different units of space and time. These carriers of motion ordinarily conform to the type of the rigid body, and among the latter predominantly to rigid bodies of the type 'actor'. Frequently space-time connections are also brought about by potential carriers of motion, who for their part are produced by an already familiar discovery of narrative cinema, namely that of the stare. The latter for its part forms a unit, indestructible in the context of narrative grammar, with the carrier of motion, which has degenerated to a rigid body, ie the rigid type of actor.

The basis of this construction is the unshakeable identity of an image with what it portrays. The image of a space is perceived as being identical with the space itself. As a result, distortions of scale are generously overlooked from shot to shot. Equally overlooked are distortions in the area of colour distribution, even to the surprising extent of black and white films being identified with a particularly emphatic kind of realism.

What are not overlooked, however, are visual errors such as extreme over-exposure and wide-angle distortion. In these cases the identity of picture and subject is undermined to such an extent that within narrative grammar they are located in the province of psychotic and sub-criminal perversion.

A curious value is given to heavily under-lit scenes to which, strange to say, the time co-ordinate 'night' is attributed even when a thousand pieces of evidence contradict this interpretation. A brownish tint with black-and-white footage is even interpreted as a record from time long past, although this brownish tint was at no point in time part of the chemical production process of film stock.

This suggests that narrative cinema and its grammar is a matter of a meagre collection of little artifices, to which the spectator reacts with Pavlovian certainty. Such an approach, however, overlooks the fact that pictorial errors belong to the border-regions of narrative grammar. They have little to do with the appreciation of space-time constructions, and the latter form the back-bone of this grammar. No analytic approach which stops at the level of pictorial deformation can proceed towards an understanding of the narrative principle.

The basis of the narrative system is the identity of picture and subject.

The systematic disregard for the questionable status of such an identity leads to a bizarre understanding of the reality of narrative film. Films which permit unambiguous space-time constructions, without the synthetic element of their process of production being apparent in them at first glance, are felt to be particularly realistic. The very films, however, that can produce this lack of ambiguity are synthetic to a very high degree, and so they must be too. They insist, for example, on the stare.

A society whose pictorial appreciation is such that it confuses pictures with reality, even when a thousand pieces of evidence weigh against the possibility of such a confusion, must in itself be or become highly schizophrenic. How then can an individual understand reality, if he concedes reality only to systematic distortion. If he tries to evaluate reality according to the standards of a deformed reality. How can he interact with other people, if his understanding of reality is determined by the reality of the stare.

B) Normalisation Towards the Representational Shot

The identity of Picture and subject calls for far-reaching restrictions within the possible forms of pictorial representation. We mentioned constraints towards standard lenses, correct exposure, and plausible colour-balance. On top of this comes the demand for satisfactory picture definition. This is necessary to prevent the identification of pictorial space with real space being undermined. Likewise a film's time in the narrative context must appear equal to the time it represents. This leads to mutually standardised speeds for shooting and projection. 24/25 images per second is the norm. Small deviations can be noticed only in sound films by an audience with sharp hearing.

A further restriction consists of the use of the static camera. The identification of picture and subject leaves no room for a character outside the scene who is making the scene come about. The cameraman must remain invisible. Pans are used only if they are so strongly motivated by an observable motion in the scene that the action of panning is not felt to be an operation by the camera. All other pans have an atmospheric character and only condense the narrative continuity produced by static camera shots. All experiments with hand-held cameras stress the doubtful nature of the identification of image and subject and only a well-disposed audience perceives them as narrative. Even when such shots are officially designated as 'especially documentary', their effect is only atmospheric. The identification of the image with its subject becomes ineffective after a time. The film is misunderstood as the product of the cameraman.

Conventional narrative cinema dies once the static shot is abandoned. And it was the static shot which made its rise possible. The first film cameras had to be used as static cameras, since there was, as a rule, no means of viewing the image while the camera was running.

The principle of identification excludes a further area of images in which

an uninterrupted identification of picture and subject is not possible in spite of existing normalisations. This area includes pictures of extremely small objects. They have to be prepared for by other shots, so that an acceptable sense of scale is established. There is moreover a range of shots in which spatial relationships are so opaque that they can only be read as abstract images. Those too are forbidden in a strictly narrative context.

Another requirement is the horizontality of a depicted horizon. If discrepancy is hot given an extraordinarily strong motivation, then it is perceived as the deliberate action of the cameraman and works as such against the principle of identification.

The horizontality of the horizon demands, even in spaces without an horizon, the verticality of verticals. The vertical is the determining structure in a narrative image. Even when, because of the selected camera-angle, the horizontal lines are distorted for purposes of perspective, the verticals remain vertical. Foreground/ background focussing must be motivated by a previous action. It is subject to the same rule as the pan.

Anything which is perceived as a marked deviation from the abovementioned normalisations has an atmospheric effect, and many of these deviations have a striking tendency towards the domain of the psychotic and sub-criminal. Sub-criminals and psychotics are accepted as people for whom the world has lost its bearings, and in narrative cinema this means in effect that the unshakeable identity postulated between image and subject is undermined.

The identity of image and subject can also be approached through the concept of the representational shot. A shot represents a clearly defined space and is identified with it. We designate as representational those shots in which there appears to be a simple identification between the image and its subject. In this sense representational shots demand the normalisations described above. The representational shot is a prerequisite of narrative grammar. A simple succession of representational shots is however, insufficient to create a narrative system. Narrative grammar -is more tightly structured.

Also some non-narrative systems use representational shots, as for example the genre of documentary film. But unless they are strongly structured in a narrative way, their effect after a while becomes atmospheric, like the illustration to a text. This is also extremely true of news broadcasts which incorporate documentary material, where an intensive use of the hand-held camera obviously increases this impression.

Representational shots must have a specific minimum length, so that their representational effect can be recognised. This minimum length lies between one and five seconds. If the shots are shorter, they have an atmospheric effect.

The representational shot is unconnected with the field of vision of a hypothetical spectator. The field of vision of a spectator does not have the least similarity with the field of vision of a representational shot. The eye sees differently from the camera. A camera-pan and the field of a person's vision as he turns his head to see more, are two fundamentally different things. Occasionally in narrative films there is an apparent identification of viewpoints between camera and spectator which is described as the subjective camera. This identification is absolutely artificial and exists only as a convention. The shot used in the narrative system usually requires no motivation in terms of a possible spectator. Camera-viewpoints least accessible to a viewer do not rule out the representational character of a shot - on the contrary, frequently they provide the primary condition of its possibility.

The use of non-representational shots is at the present time examined almost exclusively in avant-garde cinema. They are however also used in commercials, and in a certain type of musical, admittedly in their most trivially atmospheric form.

Multiple exposures and superimpositions have a narrative effect only if the character of the superimposition is disguised, as in travelling mattes and other effects. If they are recognisable as multiple exposures or superimpositions, they have an atmospheric effect. One exception is provided by the dissolve from one shot to the next. Like the fade, it has the character of a well-defined signal in the narrative context.

The constant application over many decades of these principles of normalisation has so exhausted the visual raw material of narrative films that it is no longer in a position to carry contents in tune with its times. A few directors have been trying for some time to integrate non-representational images into the narrative framework. Since they however accept the narrative principle as a corset, we find more and more in their films that category of cinema-pyschopaths which have nothing, absolutely nothing, in common with true psychopaths.


C) From Speculative Continuity to Potential Neighbourhood Relationships

If we conceal part of a representational shot, its representational character is not necessarily lost. The audience's consciousness reconstructs from the information available a large amount of the hidden part of the picture. Every narrative film trains us in this knack, since every actor conceals part of the space which contains him. If he moves, he reveals a part of it. In each film we experience a thousand times direct inframe continuity.

The continuity of a representational shot also takes place out of frame. In the first place there is a direct, comparatively integrated one, which occurs through the continuation of certain lines of perspective, of vegetation structures or through the observation of lines of movement, of carriers of motion entering or leaving the frame. But then there is a speculative out-of-frame continuity, a continuity which sees the image as part of a whole and constructs for itself a geographical and social set of surroundings out of the extract tendered. The image of some rustic room evokes a rural environment of considerable extent with larger cities on the periphery, a series of social structures, etc.

An isolated narrative film, placed in an environment which is ignorant of narrative films, is absolutely incomprehensible. A continual confrontation with its products is indispensable to the comprehension of the narrative system. Narrative cinema is serial in principle. The grammar of narrative is constituted in the spectator through the process of seeing a series of films which all obey the same grammatical laws. These frequently have a certain plausibility, frequently however they are a matter of coincidentally chosen conventions which have to be learned. We take as 'narrative' what is common to all narrative films. Narrative films moreover are films which we can see in cinemas, and by 'cinemas' in this context we do not mean those cinemas which have devoted themselves to a dubious concept of film art. It cannot be overlooked that even the films which we see there are narrative through and through. The distinctions are of the finest kind.

Speculative out-of-frame continuity differs from direct out-of-frame continuity in that it does not orient itself by frame. (Note: Direct out-of-frame continuity - you look at the edge and construct a direct (meaning close) environment to the edge. Speculative out-of-frame continuity - you look at the picture as a whole, at its real 'meaning' and from this you derive an entire environment.) It is the result of an appraisal of the spatial qualities of the image as a whole. Thus it is irrelevant for the understanding of narrative films whether a more or less considerable part of the edge of the picture is cut off in the course of projection. In fact one always notices that the screen-area in cinemas is nearly always considerably reduced. This reduction seems however in no way to impair an understanding of the films. Speculative out-of-frame continuity orients itself principally in the centre of the picture.

When the camera moves our attention turns to those edges of the frame, where most activity is taking place. This form of attention reduces the centrally-oriented speculative out-of-frame continuity. For this reason too the moving camera is as a rule used for atmospheric effects, or the movement of a carrier of motion at the centre of the picture outweighs the movement at the edge of the frame. (Note: camera movement, especially pans, generate by their nature great suspense in the area of the edge where new information passes into the image. So the attention goes to this edge. If an important carrier-of-motion is moving centre-frame then attention is centred again, the picture is grasped as a whole and speculative out-of-frame continuity can take place).

If we have two disjunctive representational shots before us, then for each there is a speculative out-of-frame continuity. According to the intersection of these two continuities, the spectator comes to a decision concerning the potential neighbourhood relationship (NR) of the two spaces shown. Arriving at such a decision is essential for the understanding of narrative films.

Because of the speculative quality of out-of-frame continuity this decision itself is only speculative. But since in narrative films judgement must always be passed very quickly., it must at the same time be crudely hierarchised. The concept of direct and distant neighbourhood relationships offers itself as a model for such a hierarchy. According to this terminology we designate two shots as being in a potentially direct neighbourhood relationship if their picture areas can be immediately juxtaposed. The decision concerning this type of potential NR takes into account the direct continuity of the two shots. If a potentially direct NR seems impossible between two shots, they can nevertheless be in a potentially indirect or distant NR according to whether or not one allows that the two picture-areas are to an extent close to each other. The notion of closeness as a spatial category is very unclear in this context and can only be more exactly ascertained through the inclusion of a time-element. In films whose chief means of transport is the car, a different concept of closeness is in play than in films whose actors have to make their way on foot. Hence we describe as being in a potentially indirect NR those shots the distance between whose picture-areas the spectator surmises to be passable in a relatively short time through the means of transport available. Everything else can be categorised as probably only distant.

These potential neighbourhood-relationships between two shots are translated into actual neighbourhood relationships through a series of clues, which one in addition infers from the images. Thus potentially direct NRs can be transformed into actually direct, actually indirect or actually distant NRs. Two rooms, for example, from which one could infer that they were in a direct NR may turn out to be situated in two different cities, and therefore in fact only distantly related. A potentially indirect NR excludes an actual direct NR, but can change into an actually distant NR. Potentially distant NRs exclude actual direct and indirect NRs, for a house in the Rocky Mountains and one in New York cannot be in actually direct or indirect NR.

A similar structure of neighbourhood relationships exists at the temporal level. Two shots can be linked by temporal continuity, or a relatively slight interval of time can pass between them, or else a considerable lapse of time may have occured between them. Here too we begin with a speculative decision arising out of an initial image analysis, which is later modified on account of a series of other indications. If a day-shot follows a night-shot we can tell that a number of hours must have elapsed in the meantime. Usually however lapses of time are not as easily noticeable and the spectator must mostly convert potentially direct temporal NRs into actually indirect or actually distant NRs.

The out-of-frame continuity of a shot is the pre-condition for the emergence of the conception of a potential NR between two shots. In its turn this conception is necessary for an understanding of the actual NRs. Understanding a narrative film means in effect understanding the spatial and temporal relationships between the different shots. It is probable that this is the only thing one really understands in narrative films. A description of the grammar of narrative is identical with a description of the spatial and temporal constructions of this grammar. These are the things which no audience can avoid appreciating, while everything else can be interpreted one way by one spectator and another by another. Research into it is speculative unless carried out in concrete detail on an empirical basis. Certainly this other area outside spatial and temporal constructions, does not have the characteristic of a grammar, ie to be understood by large numbers of people in the same way.

The aim of this essay is not to conceive of this grammar as something God-given and immutable. It is much more a matter of a system, which has in itself such critical domains that it systematically destroys itself. It is a living system with a limited life span. The more films per unit of time that are produced according to this system, the more quickly it dies. Or more circumspectly expressed, the more it functions less. This essay is written at a point in time when the grammar of narrative has already entered a coma, and for film there is only one alternative: either it dies with the demise of narrative, or it lets narrative fall away and turns to other systems of organisation, which contain the grammar of narrative at most on their peripheries. The fate of cinemas, however, seems determined in either case: at the very best they will assume the functions of museums.

D) Film History

On May 20th 1909, the premiere of Resurrection took place, starring Florence Lawrence and Arthur Johnson. Resurrection is a film by D.W. Griffith and is based on Tolstoy's novel of the same name. It is about 10 minutes in length, with the following structure:

T1 : T2 : X1 : T3 : X1 : T4 : X2 : T5 : X2 : T 6 : X3 : T7 : X4 : T8 : X4 : T9 : X4 : T10 : X5 : T11 : X5 : T12 : X5 : T13 : T14 : X5 : T15 : X6 : X7 : T16 : X7 : T17 : X7 : T18

where Xl, X2, X3 etc indicate the picture spaces of different representational shots and T1, T2, T3 various titles. The symbol ':' designates a cut.

The titles are as follows:



















The settings are:

Xl: A drawing room (appears twice)

X2: Katuscha's room (twice)

X3: A tavern (once)

X4: A police-station, also a court-room (three times) X5: The prison (four times)

X6: A snow-covered landscape with little houses ('SIBERIA') (once)

X7: A different snow-covered landscape with cross and church (three times)

An interesting feature of the narrative form of Resurrection is the fact that, until the transition XIB:X7 all the shots are joined together by titles. The standard join between the two shots Xi and Xj takes the form Xi :Tij :Xj in which Tij designates the title which intervenes between Xi and Xj.

The scenes are all in long-shot, the typical shot par excellence. We know that until about 1910 the long-shot was used almost exclusively in narrative systems. Not until more developed forms of film were other representational forms, such as the close-up, assigned a place in the narrative system.

A striking feature of Resurrection is the fact that nearly all the titles describe in advance the events of the scene that follows them. The medium seems at this point in time so uncertain of its own capabilities, that it is thought necessary to duplicate information to a certain extent, and that the images only have the value of illustrating an existing text. The main thread of the action is modelled on the chapter headings of a book unconnected to the structural relations of the film-images. The sequence of images acts as an atmospheric code to a literary sequence. Subsequently this has changed only superficially in the narrative film, particularly since the introduction of the sound film, which has increased the value of the literary code. In all narrative films the images have very much the quality of illustrations to a dialogue.

The heroine Katuscha appears in all the scenes in the film. The film is her film. The emerging spatial relationships orient themselves towards her appearance. They are supported by titles 0n a Low Tavern', 'To a Life of Hard Labour in Siberia'). Resurrection has a linear narrative pattern, which views an individual character from different points in time. This pattern has remained until this day the fundamental macroscopic element of the narrative film. All narrative films pursue the fate of one, two, or three characters at ,essential' moments in time. These characters bind the film together. When too many destinies are described in too much detail, narrative films effectively lose their narrative quality so that, as a rule, they no longer reach the cinemas. Narrative cinema makes the division between lead and supporting actors inevitable. It is an obvious assumption that the space-time constructions of films with only supporting parts become so complex that they can no longer be clearly enough understood by an audience. And not to understand the space-time constructions of a narrative film is the same thing as not understanding a narrative film.

This fact has an ideological import. A grammar which necessitates a hero can effectively only describe a story from the perspective of the hero. And any attempt to describe history as the result of a mass-movement is doomed to failure in the cinemas if not sooner. Historical films are for this reason reactionary from the start, or they are approached as costume dramas, free of value judgements. The same problematics probably affect the novel, from whose code the hero-principle seems to stem. There, of course, we are dealing with completely different conditions of production.

Another principle feature of the narrative film is the phenomenon of the compression of time. Resurrection has a duration of about 10 minutes and covers a period of about 6 years. Compared to the periods of time they depict, the duration of narrative films is usually ridiculously brief. The scenes of the film thus bear a tremendous significance, since they have to demonstrate their unique ability to depict the most important stages of a person's destiny. They become thereby doubly representational. Not only do they represent the physical scenes but also the decisive stages of a person's destiny. This representativity of content is demanded of the succession of scenes, called sequences in present-day films. The isolated single shot is now less important in narrative films compared with Resurrection.

The present tense is used consistently throughout Resurrection in the titles. It is the clearest consequence of the notion of the spatially representational shot. A depicted scene has to be perceived as an 'is'-scene and not as a 'was'-scene from the past. Everything that takes place visually has to be perceived as present. This alone ensures the identity of the image and that which it depicts. Indicative in this respect are films which are set in the past and which force one to throw oneself back into the past, as though one were on the spot, and not from a historical perspective. This leads to unbearable tensions in the images, since film makers have to take pains to inject into the historical surroundings an authenticity far beyond that which is known to us of the period in question. Clearly such efforts lead to a totally distorted understanding of history. It is interesting, however, that it frequently functions in spite of obvious contradictions. An explanation of this phenomenon can be found initially in the analysis of systematic parallel montages. A film like Resurrection, however, does not take us back to the time of Czarist Russia, but to the America of 1910. Its simple linear form with titles reduces our confidence in the representational character of its scenes.

The form of Resurrection is essentially:


This film form is the simplest device for binding together various scenes. It is at the same time the first film form to make possible a systematic, serial narrative production. It permits scenes foreign to each other to be combined in such a way that a 'meaningful' structure of action emerges. Any reference which cannot be gathered from the images can be defined by means of the titles. With this film form serial narrative film-production becomes a sector of industrial production proper.

It is no coincidence that, simultaneous with the emergence of this form, there began the first large-scale concentrations of capital in the film business. With the Motion Picture Patents Company (MPPC) the first horizontal and vertical monopoly in the history of films was established.

Industrial production was made possible by the fact that a large quantity of literary material could be translated into script form by people who needed just to be able to read and write. Because of the prevalence of long-shots anybody could be employed as an actor, and for direction one could fall back on directors with a humble measure of theatrical experience. This made the principle of low remuneration viable since almost anyone could carry out the necessary work. Production was extremely cheap, and a monopoly price policy could be operated against the cinemas. The film form as described ensured the accumulation of capital necessary for the later verbal developments which arose primarily from Griffith's work of 1909 to 1911. Although further development had to contend with the opposition of the MPPC who, as an employer's maxim, held to their policy of short films and rock-bottom wages. Banks nevertheless began to get interested in putting capital into narrative films produced as series.

This cinematic form is nowadays by no means as pass6 as one might perhaps think. Its use is still standard in the documentary film, with the slightest difference, that the titles are no longer there to be read, but - and they call it progress - they can be heard as voice-off on the soundtrack.

This form certainly need not be relegated to the past, even in as far as its efffectiveness in conveying the content of a film is concerned. I doubt whether a modern system of visual narrative, whatever its technical input and cost, could ever be as effective in conveying the content as the use of titles:


The motive behind allowing titles to disappear from films was certainly not the idea of optimising the flow of information. The narrative film in its present form is an extremely bad information system.

The scene of the action of a film produced according to the form


is potentially the whole world. In fact, however, locations remain restricted to studio decors. According to the policy of keeping film costs as low as possible, every location shot was an irresponsible luxury. Transport and wage costs, and waiting for good weather, made a single exterior shot more expensive than an entire film produced on the revolving stage principle. Furthermore, exterior shooting, unlike interior, does not have the same pleasing characteristic that characters can repeatedly meet and interestingly interact. (Note: in a room people can meet all the time, talk, have some interaction and nobody is surprised. Outdoors this is theoretically possible too, of course, but our experience in life tells us that outdoor interactions are generally dynamic, people tend not to stand too long in one place, and then other people meet in the same place etc). Often they can be used only once or twice in a film, whereas the interior scenes offer a whole reservoir of theatrical interactions. In spite of the enormous potential, the locations of narrative films produced according to this model are restricted to cheaply constructed interiors and painted exterior decors. This is broadly still true today of most narrative films.

E) Titles and Operators

Let us now examine more closely the module Xi:Xj, in which Tij is a title which joins together the two pictorial shots Xi and Xj. To this end we formally divide the title Tij into one component which describes the spatial and temporal connections between Xi and Xj, (T (ST) )ij, and into a residue (T (Residue) )ij. Tij = (T (ST) : T (Residue)) ij.

The way we imagine this is that first one part of the title is visible, then the other. As the division is artificial, we can assume that it does not depend on the sequence. The title 'FIVE YEARS LATER - IN A LOW TAVERWT6) would for instance break down as follows:

T (Residue) = a LOW tavern

The quality of the tavern, a tavern belonging to the lower strata of society is not so much a quality of space and time as of content. There are titles which have no space-time component, just as there are titles which have no 'residual' component, eg


The 'residual' portion of the titles we will hereafter describe as the content related component of the titles.

The effect of a title often reaches beyond the following shot and then becomes a formal component of a later title. In Resurrection, for example, we find the system

X5: T13 : T14: X5: T15: X6 in which X5 is meant to represent a prison and X6 a landscape in Siberia. Now T14 reads 'BUT NOW, THE POLICE ARE READY TO TAKE HER TO A LIFE OF HARD LABOUR IN SIBERIA', in anticipation of which, in X5, one sees the police taking Katuscha away. Then T15 appears = 'KATUSCHA TRIES TO HELP THE POOR AND UNFORTUNATES WHO S HAR E H ER FATE', and she is then seen in the 'Siberian' shot X6. In the space-time connection between X5 and X6 the information from T14, that we are now off to Siberia must be taken into account. Similarly, completed space-time connections must also take into account previous titles by storing them up for a time to be brought into play when the relevant shot appears.

This structure can often be seen in modern narrative films too. In the course of a piece of film dialogue we may learn, for example, that one of the main actors plans to go to Chicago in the next few weeks. Then, after a time, we see him in an environment which could represent Chicago. At that instant the space-time title stored in our heads is actualised, and we produce the reading: A FEW WEEKS LATER IN CHICAGO. This space-time connection through dialogue is not really different from the one produced by appropriately placid titles.

The developed narrative film form of today only occasionally contains titles. As a rule they are implied in a piece of dialogue. Now and again, however, titles crop up such as LYONS or THREE YEARS LATER. Frequently a visual image can also stand in for a title, such as, for example, a picture of Tower Bridge or of Big Ben with the sound of bells which is synonymous with the title LONDON. The content element of titles has been fully absorbed by the visualised system of action.

If we were to disregard the content elements of the title in our film form SHOT : TITLE : SHOT : TITLE : ... and designate the space-time component of the titles as STT (Space/time/title) then we have a version of the model


ie, a space-time title connecting together two shots and presenting the spatial and temporal connections between them.

Now the achievement of film grammar consists essentially in eliminating the explicit presentation of these space-time titles without the spectator losing his space-time orientation, so the film form moves formally to the system :

SHOT: SHOT: SHOT: SHOT: ... and in each shot there is a series of indications which mediate the spatial and temporal organisation between the shots.

As a model we can imagine these indications consolidating in the mind of the spectator to form a space-time operator, which produces the necessary connection between the spatial and temporal segments. In this terminology the analysis of narrative grammar is identical with the analysis of the presentation of the space-time operators contained within it.

If we designate a space-time operator as STO, then film form assumes the shape

SHOT: STO: SHOT: STO: SHOT: ... What we expect of these spacetime operators is that shortly after a cut they are assembled in our minds from certain indications that we can gather from the shots, in order then to mediate the spatial and temporal connections between the individual shots. Also the form

SHOT: STT: SHOT: STT: SHOT: STT: ... works according to this principle, if we accept that space-time operators also emerge from written titles.

Space-time operators are not exclusively determined by the shots by which they are immediately surrounded. As in the case of titles, indications which have cropped up earlier are explicitly able to modify a space-time operator. Likewise it is possible for a space-time operator

point to influence a later one.

The emergence of space-time operators is closely connected with analysis of the neighbourhood relationships Ms) between two shots. Potential NRs induce a potential space-time operator, which then through further indications becomes an actual space-time operator in order to transform potential NRs into actual ones. This occurs as much on the spatial as on the temporal level.

F) The Destruction of Simultaneity and its Reconstruction (Film History 2)

The images of a film massively oppose every attempt at a verbal description. The description of even a single frame is, in spite of a century-long tradition, extremely inexact and is not capable of replacing the image. And even if this were half possible through extreme simplification of the image

or lengthy attempts at description 1 the phenomenon of movement erects such high, new barriers, that speech, insofar as it claims to be a substitute for reality, must totally despair. If we have two people before us, independently carrying out different actions, the simultaneity of their actions invariably forces us into such grammatical constructions as 'While A does this, B does that', which must so accumulate that the actions of A or B can no longer be understood in terms of closed systems, but lack information about the possible interconnections between two actions. This problem is typical of the difficulties of a literature which regards itself as objective, for it appears in similar form in the description of reality, which -and such is its nature - contains a mass of simultaneous processes. The 19th century novel simplified matters in a way which even today represents the prototype of realistic writing. In Balzac we read:

'Is it that good', asked the advocate, handing Thuillier the piece of paper. 'Yes it is', he answered, folding the letter carefully and sealing the envelope. 'Now address it', he added.
The letter found its way back to Le Peyrade.

Or a page later:

When they were alone, Le Peyrade took hold of a newspaper and seemed to become absorbed in it.
Thuillier, who had begun to feel uneasy about the solution to the problem, was regretting that another plea had occured to him too late.
'Yes', he thought, 'I should have torn up the letter and not gone so far in furnishing the evidence.'

(Balzac, The Petitsbourgeois, 1843)

A realistic literary structure therefore looks like this: first one follows A for a period and then B, a process in which we only occasionally receive information about what A is doing while we watch B, and vice-versa. The technique is one of a rapid parallel-montage, in which the so-called 'essential' phases of movement are recorded and the inessential are not. This may be acceptable in a literary structure, but it is inadequate for the demands of a reconstruction of reality, for what does Le Peyrade do, after he has handed Thuillier his sheet of paper? What was Thuillier doing, while Le Peyrade was passing him the paper?


What did it actually mean when Le Peyrade took up a newspaper and seemed to bury himself in it, while Thuillier had already begun to get nervous about a certain question?

This 'realistic' construct, which exists in the modern novel only as caricature, celebrated a resurrection in narrative film. It is what determines its structure and is responsible for the curious and erroneous belief that narrative films are often 'more realistic' than reality itself.

Its basic prerequisite is the complete destruction of the confusing simultaneity of the single shot. The more clearly it is destroyed, the more credible appears its rebirth in the 'realistic' construct. Here we discover one of the roots of the degeneration of the actor to the rigid body with limited options for movement.

In what remains to us of Méliès' 1905 film "From Paris to Monte Carlo*" (*Also known as An Adventurous Automobile Trip; see Harnmond, Marvellous Méliès, London, Gordon Fraser, 1975) we can see the principle behind this destruction. Projected at twenty-four frames per second the film lasts 5 minutes and 20 seconds, and, if we take the individual scenes to be realistic in themselves, has twelve shots, which each appear only once. In nearly all these shots we see very many more than just two or three characters engaged in activities which are not amenable to description. In each shot there appears after a certain interval the chief protagonist in the film, a red-coloured pasteboard car continually moving from right to left across the picture against a black and white background. The story of the film is the story of the obstacles which stand in the way of the red monster as it makes its way from right to left. Whatever activities occur in its path are, with fatal consequences, crashed into, run over, and destroyed by it. For the car has to make its way from right to left in order to be able to appear in the next shot. The activities in the picture, however, are doomed even before their physical destruction. Any simultaneous human activities in the individual shots are obliterated as soon as the car appears in the picture. The eye immediately shifts its attention to this object, so sharply distinct from its environment. All other activities are reduced to mere decoration, and, if they do not explicitly support the movement of the hero, are reduced to the function of being obliterated by it in the most physically spectacular way. The hero conquers the complex simultaneity of reality and along with it reality itself, first in the viewing eye and finally physically.

This is the message of narrative film to this day.

Actors in film therefore not only have the property of concealing the space behind them and thereby training our eyes for the in-frame continuity of a shot, they also tend to help solve the problem of simultaneity in the individual shot. When the main actor appears, attention is directed towards him, irrespective of whether he looms large in the image or is only a tiny point in a long-shot. This is a reflex which is closely coupled to the construction of narrative grammar. When the image appears before the spectator, it is immediately hierarchised and scrutinised to see if we can discover a main actor in it. Only when this question is answered can attention turn to other parts of the image. For the main actor usually produces such intense activity that background details are submerged by it and an analysis of the image cannot be undertaken, or only at the risk of one being unable to understand subsequent actions.

This mode of viewing images has even become embodied within the systems of production for narrative films. It led to the star-system and to extremely high salaries for actors. In many films we observe that the film only truly begins when the famous star appears. Everything that takes place up to this point is preparatory skirmishing and may be happily overlooked. The first entrance of a film star marks the commencement of a narrative film.

This, of course, has an intimate connection with certain areas of bourgeois ideology which have not remained hidden. There have always been social critical analyses of the star system. What has been overlooked though is the i-ntimate connection which this ideology forms with the narrative film-form itself. The 'main actor' principle is an important precondition for the narrative preparation of sequences of events, and only the mutilation of images brought about by this principle makes the sequential presentation of simultaneity in narrative films possible.

This way of seeing images is by no means the only way in which one can see images. In fact it is a way which in actuality only occasionally takes place, perhaps when one unexpectedly meets an acquaintance. Even so this acquaintance becomes after a time a peripheral phenomenon amongst all the other peripheral phenomena which are attendant upon one. In narrative film however this way of seeing is constantly reinforced and is indispensable for an understanding of its space-time constructions.

In the realistic novel of the 19th century the imperfection of language enforces an abstract, simplified presentation of the actions which are described, which is further reinforced by the temporally one-dimensional nature of writing. In film, whose capacities are by contrast towards the representation of simultaneity, the structures of the realistic novel were imitated, and the presentation of simultaneity was forced into the same temporally dimensional model that had been provided by literature with its limitation to the written word. This is how it became possible to examine and describe films according to the 19th century literary code. Even film criticism describes narrative films as if they were 19th century novels. And the fact that critics, trained on such products, can find very little to say about films with non-representational shots, is not particularly surprising.

G) Systems: Simple Linear Cuts, Return Cuts, Transitional Forms, Atmospheric Cuts

We now wish to dissect analytically the many and various editing forms of narrative cinema. The first analysis proceeds from the concept of the disjunctive cut. A disjunctive cut is a cut in which the image-fields of two successive shots do not overlap. On this basis we can say that the cuts to be found in narrative cinema can be divided into those which are disjunctive and those which are not. Disjunctive cuts moreover can be divided into simple linear cuts, return cuts, a form of transition between the two and into atmospheric cuts. Simple linear cuts are those in which the same carriers of motion, for example actors, can be seen before and after a cut in two different disjunctive shots, where the shot following the cut is as yet unknown to the observer at the point of the cut. Return cuts are cuts which connect shots already known to the observer at the point of the cut. We can divide them into linear and non-linear return cuts, according to whether, following the return-cut, a carrier of motion passes from one shot into the next. The forms of transition break down into retarded or postponed linear cuts and advanced or anticipatory return cuts. A retarded linear cut is a simple linear cut which is modified in such a way that between shots linked by a carrier of motion one or more other shots are inserted. Hence too the term delayed linear cut. By advanced return cut, on the other hand, we mean a cut to a shot followed, after the insertion of one or more intermediate shots, by another cut to the same shot. which we could have assumed to be a simple or retarded linear cut if we had failed to take the advanced return cut into account. Therefore it is also termed an anticipatory return cut. Atmospheric cuts, finally, are disjunctive cuts which are not embraced by the preceding classification. Non-disjunctive cuts are also called overlapping cuts.

The analysis of narrative cuts is complete and can be diagrammatically represented as follows:


 Narrative cuts

 Overlapping cuts

Disjunctive cuts 


 Simple linear cuts

 Transitional forms

 Return cuts

 Atmospheric cuts


 Retarded linear cuts

Advanced return cuts

 Linear return cuts

 Non-linear return cuts



When we observe that in a narrative film different spaces are linked by the lines of sight of different actors, then we attribute to each of these lines carriers of motion which we cannot see, but which we however can imagine. We call them virtual carriers of motion. Thus cuts which produce spatial connections through lines of sight fall within the category of simple linear cuts or into the category of the advanced or linear return cuts.

Simple linear cuts and transitional forms are space-constructing cuts in the sense that they mediate spatial relationships between the individual shots. Return cuts are cuts which use space, which work on space construction. Both forms of cutting simultaneously generate a temporal structure. This structure is highly dependent on the exchange of carriers of motion between individual shots. Atmospheric cuts are only significantly involved in this spatial and temporal construction when they explicitly produce a title indicating a shift in space or time. We have already encountered such a system when we conferred upon a picture of Tower Bridge in the narrative context the title 'London'.

If we imagine various overlapping shots being joined with one representational shot, narrative films have a quite distinct form. A sequence of space-constructing cuts is followed by a sequence of return-cuts, which is further followed by a sequence of space-constructing cuts with following return cuts, and so on. This system is interrupted only by occasional atmospheric cuts. It is to be observed that in narrative films return cuts are by far the majority of the forms of cuts encountered. Films are rarely seen in our cinemas which do not obey the dominance of the return cut. This naturally does not mean that any film which largely consists of return-cuts will appear in the cinemas for this reason alone. And it also does not mean that there have not been periods in the cinema when films did not obey the dominance of the return-cut.

The return-cut is the back-bone of narrative grammar and at once both its strongest and weakest point. Its strongest because its systematic application produces what is generally described as filmic suspense. Its weakest and at the same time most critical point, since in it is most strongly revealed what one can define as the postulate of the universal presence of the spectator, which is, moreover, the principle that we will discover to be responsible for the current collapse of the system of narrative cinema as a whole.

Among the many return cuts which characterise the narrative film, those which are generated through the mutual gaze of two actors are by far the most common. These looks belong without exception to the category of the stare. If one reduces feature length films by a time-lapse process to a length of one or two minutes, whereby each shot is represented by one or two frames, the impression that the problem of narrative cinema consists in achieving an exchange of looks between two people is inescapable when one views the abridged versions. And any film which negates this system loses the narrative quality which enables it to be shown in the cinemas.

Previous attempts to describe the grammar of narrative cinema always started from an analysis of the atmospheric cut, extrapolated from the works of Eisenstein and Pudovkin and assumed that modern narrative films were still constructed on their general principles. What was overlooked in the process was the fact that the meaning of the atmospheric cut, still vital in the case of Eisenstein and Pudovkin, was severed at the root by the introduction of the sound film. The narrative cinema of today contains the atmospheric cut as no more than a cabbalistic accessory, and only then in its most trivial forms.

The full extent of the irony is that Eisenstein's silent films must appear well nigh incomprehensible to a modern audience and can only now occasionally be seen thanks to a highly developed film culture. They find themselves in the curious vicinity of modern avant-garde film, which also cannot find a wide audience, since on principle it does not obey existing grammar.

(To be continued ... )

Translation by Philip Drummond, edited by Barrie Ellis-Jones with additional material by Elizabeth Reddish.

Nicht Geordnete Notizen zum Konventionellen Narrativen Film. First published in Boa Vista 1975 and later in Filmkritik 274, October 1979.

back to English summary | zurück zu Aufsätze