Filmmakers used the short story, drama, and novel as sources for new conceptions of causal and temporal unity. At the same time, they needed to find means of creating a unified spatial structure in which story events could take place. To a limited extent, these means could also come from the other arts: the film frame is analogous to the proscenium of a stage; the sudden leaps of time and space achieved by editing resemble the freedom of movement enjoyed by the novel's narrator.

But again, the filmic medium imposed its own unique demands. The filmmaker juxtaposes a series of disparate spaces, building from them an overall narrative space. That space is concrete, not the verbal construction of the novelist. Hence the filmmaker must be able to g uide the spectator's understanding of spatial relations if the film's causal actions are to be clear. Editing, with its instantaneous changes of vantage point, presented a new aesthetic challenge, and filmmakers had to a considerable degree to thrash out their problems on their own. The resulting guidelines became the continuity system.

The formulation of the continuity editing system was not a direct development of devices from the primitive period. Classical films of the teens did not involve more complex, more correct usage of the same devices available to earlier filmmakers. Rather, fundamental changes in the systems of causality, time, and space brought about a profoundly new approach to filmmaking, and the functions of individual stylistic devices changed as well. Some techniques used infrequently in the primitive period - the cut-in, point-of-view and eyeline structures, dialogue inter-titles, and so on - became central means of constructing space and conveying narrative information in the teens. Key devices of the primitive cinema changed in their function. The lengthy tableau shots became establishing shots; expository titles which formerly anticipated action now merely established situations. The classical system is a change from the earlier one, not necessarily an improvement upon it.

With the growing emphasis on calculating how a film could be understood came new normative systems within the production sector. The growth of a trade press after 1907 (brought about by the nickelodeon boom and regularization of production) contributed to making these norms uniform across the industry. Almost from the beginning trade papers and instruction books emphasized a specific conception of what constitutes a good film. In one of the earlier scenario books, Herbert Case Hoagland, of Pathd Freres, gave this advice to writers:2

Let one scene [shot] lead into the next scene wherever possible. Motion picture theater goers don't yearn for mental gymnastics and shouldn't be kept guessing as to who the characters are or why they are in the story at all.... Keep your scenes in a sequence easily followed by the onlooker.

Increasingly, the conception of quality in films came to be bound up with the term 'continuity.' 'Continuity' stood for the smoothly flowing narrative, with its technique constantly in the service of the causal chain, yet always effacing itself Later, 'continuity' came specifically to refer to a set of guidelines for cutting shots together, but the original implications of the term lingered on. The 'continuity system' still connotes a set of goals and principles which underlie the entire classical filmmaking system.

One of the best descriptions of continuity was written before the term was being applied commonly to film. In 1910, a commentator in The Nickelodeon outlined what constituted great films: 3

The basic principles of Hollywood film practice are here already: the story as the basis of the film, the technique as an 'indiscernible thread,' the audience as controlled and comprehending, and complete closure as the end of all. Moreover, these ideas soon came to be accepted as a set of truisms. This remark might have appeared in virtually the same terms at any point in Hollywood's history since 1910.

This is not to say that the continuity system was conceived of by 1910. Most of its principles were set forth and tested in the years up to 1917. But given this set of goals for narrative filmmaking, each new technique or device could easily take its place within an overriding formal system. The term 'continuity' itself soon came into Common usage. Initially it occurred in the scenario columns and books. Filmmakers assumed that if a scenario were correctly constructed, shot by shot, they could simply follow it literally in production, and their film would automatically have a continuously coherent narrative. So, until the late teens, references to continuity' are usually addressed to scenario writers and refer to a flow of story across changing shots. Compare

these bits of advice:'

[From a 1912 Photoplay scenario columnJ Continuity of events is a feature of the best pictures ever made. Avoid these 'twenty-year after' stories.

[A 1913 definition of 'continuity of action'J

While unity of action is one of the fundamentals of a model dramatic action, this unity must be

visibly continuous to render it distinctly and

easily perceptible.

[A 1914 scenario book:]

Unbroken continuity or perfect cohesion of story unity - of which every intelligent audience is ever conscious - that knows no such things as gaps, breaks or retrogressive movement.

[And a 1917 essay on the topic in Photoplay: ] It is the placing of the many scenes that go to make up the photoplay in a logical sequence, so that the play may run perfectly smoothly,

without breaks or jumps which otherwise would have to be covered by wordy and explanatory

subtitles.

'Continuity' quickly developed from a general notion of narrative unity to the more specific conception of a story told in visual terms and continuing unbroken, spatially and temporally, from shot to shot. This led to the word's being applied to the shooting script itself. The continuity was a numbered list of shots used as a means of planning the entire production. Thus the shot became not a material unit but a narrative one (as evidenced by the almost universal use of the word 'scene' for a ,shot'). The implication here is that filmmakers took the narrative of a film to equal the sum total of all its shots. This procedure of decoupage precludes any notion of using segments of time and space for their own sake, of elevating them above the narrative at any point. A scenarist at Hoffman-Foursquare Pictures described good continuity scripting in 1917: 'No scene [shot] which does not advance the action can be allowed to have a place in the script. Every scene must be in its proper place." Most of the rules or guidelines that were gradually formulated during the teens had as their common purpose the subordination of devices to a dominant narrative Not just shots, but everything, had to serve its narrative function. In 1914, Phillips wrote: 'We employ nothing - property, actor, scene [shots], spectacle, spoken word, insert, incident or device - in the perfect photoplay that has not a bearing on the climax of the play." In this chapter, we will see how editing rules were introduced during the teens and assimilated into the dominant filmmaking system.

The- long framing was the earliest device for creating and maintaining a clear narrative space. When other spatial devices were introduced -cut-ins, multiple spaces - the long shot ceased to present virtually all the action. Instead, it acquired a more specific function, that of establishing space. (The long shot can also have other functions, such as displaying spectacular mise-en scene or suggesting a character's isolation in a vast space, but these usually occur in addition to the basic establishing function.) Multiple spaces involve cutting together shots that show entirely different locales, whether at a distance from each other or contiguous; analytical editing cuts to portions of a single space. In the classical system, the establishing shot is so important that these other devices usually are organized around it. A film can have multiple spaces without analytical editing, and vice-versa; but to maintain 'correct' continuity according to the classical system, both multiple-space cutting and analytical editing depend on establishing shots.

The earliest staged films of 1893 and several years thereafter were tiny scenes, single events hardly long enough to be narratives. Edison Kinetoscope films usually ran less than a minute; one 1893 Kinetoscope film shows a drunken man in medium-long shot, stumbling in a park; a policeman approaches and they struggle. In such films, there is automatic narrative continuity -one event entirely visible throughout.

With the increasing length of films, an extended narrative action might be played out, still within a single locale; historians have termed this one-shot scene the tableau A one-shot film, Street Car Chivalry (1903, Edison), for example, shows a row of men sitting in a street car; they move to accommodate a pretty woman, then refuse to do the same for a homely woman until she tricks them Here we have several events forming a brief narrative, but still played in long shot within a single space; without cuts, shifts in space and time do not occur, and hence continuity is not yet a factor.

But occasional films in the years 1897 to 1903 introduced multiple spaces. In most cases, such films string a series of tableaux together, with each scene acted out completely within the space of the image and without any movement of the action into a contiguous space. In some cases, each is preceded by an inter-title, as in the most famous example of this type, Porter's Uncle Tom's Cabin (Edwin S. Porter, 1903, Edison). There was no clear-cut progression from one technique to another during the primitive period. Single-shot films (Street Car Chivalry) and series of tableaux (Uncle Tom's Cabin) continued to exist side by side, lingering into the phase of films involving multiple spaces and cut-ins.10

Later, in the teens, when scenes were regularly cut up into multiple shots, the single long shot showing initial spatial relations became one portion of the scene, usually coming at the beginning. Its function then became specifically to establish a whole space which was then cut into segments or juxtaposed with long shots of other spaces. Early cut-ins to closer framings were rare and always came after longer views of the same space and before a return to the same long view -a re-establishing shot. There was thus little need for filmmakers to specify the placement and function of the establishing shot. But in the mid-teens, close-ups were becoming standard: only a third of the ES films from 1912 had cut-ins, and only one had two of them; of the 1913 films, slightly over half had cut-ins (several with two or three, and a couple with cut-ins involving a distinct change of angle); by 1914, every ES film had at least one cut-in.

Now that filmmakers were regularly dealing with more than one shot per scene, they formulated guidelines specifying the placement of the establishing shot, cut-in, and re-establishing shot. Sargent commented in 1914 on the increasing use of close-ups: 'Lately we saw a subject in which a setting room was used. At various times three portions of this room were used for close-up pictures, instead of always using the full set.' Sargent approved the close view, but cautioned that 'it should be used sparingly, where the close-up is but a part of a scene, the opening and closing of which uses the full stage."' This advice suggests that filmmakers in the early teens still thought of the long shot as the basis of the scene, with the cut-in an occasional, effective variant.

But in the second half of the teens, Hollywood's discourse sometimes assigned a more limited, specific function to the long view. Now it became a part of an overall scene consisting of many shots, and it served to establish spatial relationships. A 1918 trade-paper review of Lois Weber's For Husbands Only recommends 'a long shot placing the locations of the various situations during the time Miss Harris overhears the conversation

between Cody and Miss Kirkwood. In the four years between Sargent's statement and this review, the conception of cutting had changed considerably. Earlier, a scene consisted of long shots, book-ending one or more close shots. After 1917, filmmakers would build scenes up from a variety of different angles, with the long shot often no more important than any other. Around 1920, Hollywood usage dubbed the long shot's function as that of 'establishing' characters' relations in space. 13*

In the later teens, filmmakers occasionally delayed the establishing shot for other purposes. A scene may mislead the spectator for comic effect, as in the second sequence of Wild and Woolly (John Emerson, 1917, The Douglas Fairbanks Corp.). The scene begins with a medium shot of Jeff seated in cowboy clothes by a teepee; a track back to long shot shows us that the "campsite" is actually inside his bedroom. Here we get not only a clear view of a new character, but the delayed establishing view humorously undercuts our first impression. Another function of beginning on a close shot is to emphasize important details which reveal the narrative situation of the scene more clearly than a long shot would. In all these functions of the delayed long shot, compositional motivation justifies the use of the less predictable schema; the variation on the standard opening is not arbitrary.

Many films of the 1920s make subtle use of the delayed long shot. The opening of Mantrap (Victor Fleming, 1926, Famous Players-Lasky) provides an example of a quick revelation of narrative situation through detail. Without any introductory inter-title, the credits lead directly to a medium close-up of a woman's foot touching a man's, which he moves away (see fig 16.1); a tilt up shows the woman speaking to a point off right front (fig 16.2). The dialogue title that follows gives the situation: '- and he said I flirted. A clever lawyer like you should get me heaps of alimonv!' After the title, we see a medium closeup of the woman, who raises a make-up case to cover her face (fig 16.3). The next shot shows her point-of-view of her own face in the mirror, which she then lowers to reveal the lawyer, with an annoyed expression on his face (fig 16.4); we later discover that his experience in handling divorces has led him to mistrust all women. Finally a tight long shot establishes them at his desk, with law books prominently visible to confirm that this is a lawyer's office (fig 16.5). The delayed establishing shot, while not the most probable schema, would remain a common alternative to the analytical breakdown of the scene.

Henry Albert Phillips, 1914

In primitive films, cut-ins occurred rarely and served a number of different functions. Most frequently, the move to a closer framing allowed the viewer to see facial expression more clearly,15 Another common function for the closer shot would be the revelation of a detail not sufficiently visible in the main tableau shot. But the cut-in could also simulate the point-of-view of a character within the scene, and occasionally it aided in the creation of a trick photographic effect.

Some of the closer shots to show facial expression were not, strictly speaking, cut-ins. Following the lead of The Great Train Robbery, quite a number of films of 1903-5 begin or end with medium shots of the characters; these may introduce the characters before the action proper begins, as in The Widow and the Only Man (1904, AM&B), where we see the two title characters in separate shots posing against a white background. The Bold Bank Robbery (1904, Lubin) begins similarly with a medium shot of the three smiling robbers in evening dress and ends with a cut-in within a prison scene; now the three are in convicts' stripes, frowning. Here the close shots structure the beginning and ending, providing a 'crime doesn't pay' moral for the whole. Some close shots for facial expression may constitute the entire action of a brief film, with no long shots to frame them, as in The May, Irwin - John C. Rice Kiss (1896, Edison) and The Old Maid in a Drawing Room (1900, Edison). So the close shot for facial expression could either comprise a whole scene or come before a longer shot of the same action.

From its earliest occurrence until the early teens, the cut-in for detail comes between two long shots taken from the same camera set-up. Barry Salt has pointed out an early cut-in in The Sick Kitten (1903?), which he identifies as a rerelease of The Little Doctor, a 1901 British film.'6 (Urban's 1903 catalogue in fact describes The Sick Kitten as an abridged version of The Little Doctor, offering both versions for sale. The Little Doctor, possibly originally entitled The Little Doctor and the Sick Kitten, was apparently made c. 1901. Only the shorter version is known to survive, but the Urban catalogue specifically mentions the cut to a closer view in The Little Doctor.) This brief film begins with a medium-long shot of two children preparing to administer a dose from a bottle marked 'Fisik' to a kitten sitting in the girl's lap (see fig 16.6). The cut-in to a medium close-up of the cat (fig 16.7) shows clearly the action of the cat lapping at the spoon's contents. Such small actions would have been indiscernible in the original framing. The Sick Kitten ends after a cut back to the medium-long-shot framing. A similar pattern occurs in The Gay Shoe Clerk (Porter, 1903, Edison), in which the central medium close-up emphasizes the detail of the customer raising her skirt to reveal her ankle; the cut-in thus explains to the audience why the clerk impulsively kisses her in the third shot, a reestablishing view of the store. In both these cases, the motivation for the cut-in is compositional, for without the closer view, we could not follow the action adequately.

Some early films motivate cut-ins as subjective shots. The subjective shot almost invariably is at least partly motivated realistically, since the camera lens is assumed to be imitating what a character's eye would see. In Grandpa's Reading Glass (1902, AM&B), a series of long shots shows some children examining objects with a magnifying glass. These shots alternate with close framings, masked as if from the children's point-of-view through the glass, of unmoving people or objects. In The 100-to-One Shot (1906, Vitagraph), there is a long shot of a grassy area in which horses are being walked before a race; the hero enters and finds a paper dropped by a rich bettor. A cut to a medium close-up, point-of-view shot shows his hands unfolding the paper.

The earliest examples of point-of-view cut-ins occur in films which depend almost entirely upon the novelty effect of the close view. Grandpa's Reading Glass contains no other action and minimal causal progression; the whole thing consists of the children's series of examinations of objects and people The cut-ins are motivated realistically (the children would see the objects from these points in space) and artistically (the close views are of interest in themselves), but not compositionally (they give us no new story information). But The 100-to-One Shot embeds its subjective shot within a larger narrative chain, motivating it compositionally by giving it causal functions; the paper in figure 16.9 contains a tip on a horse, which the hero reads and uses to win his bet. This compositionally motivated point-of-view cut-in later becomes the norm in the transitional period 1909-17.

Besides enlarging facial expression, providing details, and representing optical subjectivity, cut ins during this period could construct a more limited space within which special effects could be created. Two American Mutoscope and Biograph films which use extended and intricate pixillation shots are The Tired Tailor's Dream (1907) and The Sculptor's Nightmare (1908). In each, the basic space of the scene is established, then cut ins eliminate the human figures in order to facilitate the lengthy stop-motion process of animating objects. Vitagraph's Princess Nicotine (J. Stuart Blackton, 1909) cuts in numerous times to tiny figures cavorting on a table. Here the special effects were mainly accomplished by building over-sized matches and cigarettes, with actresses playing the parts of the princess and her friend. When trick films declined after about 1909, so did the use of cut-ins for this purpose. But certain special effects would always depend on cutting to a new view of the scene's space.

On the whole, however, before 1911 or so, cut ins were not common for any purpose. Even when closer shots became more acceptable, most filmmakers initially sought to avoid cuts within a space. Frequently staging could render a cut-in unnecessary. If a filmmaker wanted to insure that facial expression was visible, the actors simply moved closer to the camera. In the first shot of After One Hundred Years (1911, Selig) a group of characters stand outside an inn; the innkeeper comes out to greet them. One man and the innkeeper come forward to talk, thereby identifying the film's central character. This practice contrasts sharply with the more decentered framings of earlier years; compare the stock market shot from A Corner in Wheat (Griffith, 1909, AB) in which the bustling characters all claim equal attention. No framing or staging device guides the spectator's eyes to the most relevant actions in that shot. Similar framings with the characters stepping forward occur frequently in the early teens. In Cinderella (1911, Thanhouser), the Prince picks up the lost slipper in long shot, carries it forward, and extends it toward the camera, then goes back up the palace steps. A Tale of Two Cities (William Humphrey, 1911, Vitagraph) opens with a long shot of a party at the Marquis's home, with the Marquis in the depth of the shot. He comes into medium-long shot to give orders to a servant. At Old Fort Dearborn (1912, Bison '101') contains a more elaborate example, with a long shot framing a soldier who comes out of a saloon and accosts an Indian woman. As an officer and the woman's father enter to stop him, the whole group moves forward. After the Indians go out, the two men take one more step forward into plan am&icain as the officer berates the soldier.

The practice of moving characters toward the camera has never entirely disappeared. But by the mid-teens the cut-in had become an equally important way of providing a closer view of the characters. At the same period, movement toward the camera was handled in a less obvious fashion, with a deeper setting extending the acting space forward. Movement toward the camera became part of the realistically motivated staging of the scene, rather than a movement made solely to allow the spectator a better view. In The Cheat (Cecil B. DeMille, 1915, Lasky), for example, the husband and his friend stroll slowly and casually forward in the parlor after dinner. Here the movement is motivated by their desire to discuss their investment plans out of the earshot of guests in the depth of the room. There is no sense here, as there is in the earlier examples just described, of the actors crossing an empty foreground space simply to get closer to the camera. As with other devices, realistic and compositional motivation combine to make the mechanics of film style less noticeable.

In spite of attempts to use staging to avoid cutins, after 1910 filmmakers increased their dependence on closer shots. Sometimes the narrative situation necessitated a view of a detail which for some reason could not be brought forward to the camera. In the first scene of After One Hundred Years, as we have seen, the actors move into closer view. Later, in the last scene, the hero discovers a bullet-hole in the mantelpiece of his inn room. As he inspects it, there is a cut-in with a match on action. Shamus O'Brien (Otis Turner, 1912, Imp) contains a plan am&icain in which the fugitive Shamus's family read a letter from him. Barely visible outside the slatted window is listening a treacherous neighbor who will turn Shamus in. There follows a close-up of the neighbor, partly in order to catch his gleeful expression, but primarily to guarantee that we see this important bit of narrative information, which is partly hidden by the window in the long shot. In each case, the cut-in emphasizes a detail associated with a fixed portion of the set.

The Shamus O'Brien example shows how set construction could also necessitate a change of angle at the cut. But once the cut-in had come into general use, filmmakers did not always need such a pretext to vary the vantage point on the scene. The Girl of the Cabaret (1913, Thanhouser) establishes the hero, at foreground right, sitting at a table watching a cabaret violinist. A cut-in catches his reaction, moving nearly 180' to the other side of the table from the long shot. But most cut-ins still moved straight in to capture detail.

By the mid-teens, cut-ins routinely function not only to guarantee the visibility of narrative action, but to aid characterization as well. We saw in the previous chapter how the new facial acting style encouraged closer framings. This style may have been a major cause of the steadily increasing cutting tempo during the teens. Cecil B. DeMille, commenting in 1923 on the increase in the number of shots over the past ten years, attributed it to an increasing emphasis on character psychology: 17

A more regular use of crosscutting and contiguous spaces would also tend to increase the average number of shots per reel during this period. Along with the acting shift, the rise of the star system also encouraged the use of closer framings. Filmmakers moved in upon the famous faces in order to allow spectators to gaze upon their favourites. These closer shots were not the lingering glamor shots of the twenties and thirties, but they served somewhat the same function. Pickford's first appearance in her early feature The Eagle's Mate (James Kirkwood, 1914, Famous Players) epitomizes this usage. We see her first in a medium-long shot, emerging from the forest (see fig 16.23). Even though she is clearly recognizable from this view, a cut-in to a medium shot follows (fig 16.24). Here Mary plays with a bird on a branch - an action which helps to characterize her, but which also allows the camera to dwell on her. These actions could have been handled in one shot, but the division into two prolongs Pickford's entrance.

[19141 The close-view has no rival for breaking dangerously long scenes in a manner so natural and potential that oftentimes it makes a brilliant presentation of something that would in all probability have become tedious.

[19171 Main scenes must not be too long. If they threaten to be so, they must be broken up by close-ups or flash-backs [i.e., cutaways].

The cut-in thus contributed to the increasing tempo of editing in the teens. But after 1917, most writers advocated the use of close shots for specific narrative purposes - not just to liven up a scene's rhythm. A 1921 scenario manual echoed the earlier writers, saying: 'Occasionally the close-up is used to "break up" a sequence that would be too long and monotonous were the action therein contained shown in one lengthy and sustained long-shot'; however, the same writer specified that the usual uses of the close-up are 'to show a close, detailed view of that which is not sufficiently clear or which lacks emphasis in a more distant and general scene. In the case of a human face, it is occasionally necessary or desirous to show the details of expression, conveying an emotion."9

The introduction of the cut-in as a standard device and the resulting breakdown of a single scene into multiple shots bring up the question of screen direction (later to be called the 'axis of action' or '180' rule'). The maintenance of screen direction from shot to shot is one of the basic principles which American filmmakers would use to orient the spectator to the story action. There never was a period in the history of the US cinema when screen direction was random. Originally the tableau staging and framing precluded the need for any question of direction; space was presented whole Furthermore, early cut-ins failed to disturb the clarity of this space. There was seldom any question of moving to the other side of the action. The standard painted sets had only a backdrop and perhaps two small segments of other walls at the sides. In order to keep the setting in the background of the closer shot, the camera had to stay on the same side of the characters. Since filmmakers usually did close-ups directly after the long shots, by simply carrying their cameras forward, problems seldom arose, even in exteriors done without sets.

Occasionally, in later films, there are closer shots, especially of characters, taken from the side of the action opposite to the vantage of the establishing shot. Such breaks in continuity are rare, and probably result from successive shots ,being done at different times from the long shot. (We have seen how shooting 'out of continuity' was necessary to maintain efficiency in the production process.) In Girl Shy (Fred Neumayer, 1924, Harold Lloyd Corp.), close-ups of Harold's typewriter after each fantasy scene, a close-up of the villain stroking the maid's hand, and a medium shot of Harold pulling the lever to dump a workman from the back of a wagon, are all filmed from the opposite side to the establishing shot. Only one of these disjunctive cuts is compositionally motivated: since the villain conceals his gesture with his hat, we can see it only because the camera crosses over the axis of action. In the other two cases, we must assume either that the staff confused the direction when making the close shots or were willing to overlook a few irregularities On the whole, however, violations of screen direction were so rare that contemporary writers did not refer to screen direction as a problematic aspect of cutting in, but only in relation to scenes of multiple contiguous spaces.

There can be little doubt that the concept of screen direction stems from the primitive period, when the spectator viewed the action from a distance, as if in a theater seat. In a play performed on a proscenium-arch stage, one does not suddenly see the action from the other side; stage right and left remain consistent Later, as analytical editing became more common, the film spectator ceased to see the bulk of the action from a fixed point. The shifts created at the cuts by the narration do not imply that filmmakers conceived of the spectator as a disembodied spirit capable of moving anywhere within the space.

Analytical editing, Hollywood commentators tell us, follows the 'natural attention' of the spectator. First the onlooker surveys the scene (establishing shot); as the action continues, he or she focuses upon a detail (cut-in), or glances back and forth at the participants in a conversation (shot/reverse shot), or glances to the side when distracted by a sound or motion (cutaway). But while the attention may flit here and there, it never departs from the physical ties of the spectator to the degree that it crosses the line to view the opposite side of the action. Arbitrary as this conception of the spectator is, it has governed Hollywood practice from the earliest years.

By 1917, analytical editing was used consistently through whole films.

Frank Atkinson (cutter at Universal), 1924

When a film presents contiguous spaces in separate shots, it needs some method for showing the viewer that these spaces are indeed next to each other. There are different ways of providing cues: a character or object moving from one space to another might link them together; or a character looking offiscreen in one direction might lead the viewer to surmise that the next shot shows the space that character sees.

Character movement was the most common cue for linking contiguous spaces in the early cinema; it appeared widely from about 1903. In The Somnambulist (1903, AM&B), we see the interior of a bedroom and three shots of various parts of a rooftop outside as a woman gets up and sleepwalks across the roof. There is a match on the action of her coming through the door that gives a strong cue for contiguous spaces; at the other cuts, she exits to and enters from offscreen. A Search for Evidence (1903, AM&B) presents several segments of a hotel corridor, shifting laterally two doors at a time, as a woman and man move through from right to left, peeking through each keyhole. The Great Train Robbery (Porter, 1903, Edison) contains a pair of shots in which the robbers move from their hold-up of the passengers by the side of the train to the engine, in which they make their escape.

The year 1903 also saw the release of some early chase films. The chase, with its characters moving from shot to shot, became one of the standard ways of using multiple spaces in subsequent years. The Pickpocket (1903, AM&B) begins with a long shot as a thief robs and beats a man. The thief runs out at the foreground. Other people run in to help the victim; then this group runs out at the foreground as well. There follow thirteen more shots of various spaces, with characters in different combinations running through, usually toward the front and from left to right (the main directional variation being that sometimes they exit to the right of the camera, sometimes to the left). The multiple spaces even include locales above and below each other as several policemen chase the pickpocket over high stacks of lumber.

As with analytical editing, the conception of screen direction between contiguous spaces probably derives in part from the fixed position of the spectator in proscenium theater. In the later decades of the nineteenth century, spectacular productions sometimes employed a series of perspective backdrops to change locales quickly; at times, characters might move across the stage repeatedly through several represented settings, suggesting a progress through contiguous spaces. For instance, Nicholas Vardac describes an 1887 production of David Copperfield which staged the famous shipwreck scene in which Steerforth dies A series of three settings, all changed without the curtain being closed, moved the characters through space: beginning in 'The Ark Interior,'as the characters rush out to the rescue, moving to ,an area 'Near the Beach' as the rescue party runs across the stage, and finally to 'the Sea in a Storm' with the ship sinking and the characters rushing in to attempt the rescue. According to Vardac, the changes of backdrop were accompanied by 'sound effects, offstage noises, and the action itself running continuously."' For Vardac, the three scenes resemble a series of film shots, with 'dissolves' between. But beyond this, one significant aspect of this staging is the fact that the actors run across the stage to the left, exit, run across toward the left again, and so on. Were they not to maintain this constant direction, it is not clear that the spatial relationships among the Ark, Beach, and Storm backdrops would be apparent. There is but a small step from such a series of contiguous spaces on the Victorian stage to the series of shots in a primitive chase film.

The early use of screen direction depended on the fact that there were few shots, and that the same framing seldom recurred elsewhere in the film. But filming numerous shots which would later have to fit together with other shots done in different locales, on different days, made screen direction harder to control. After 1909, with the introduction of the shooting script and its attendant scene plot, it became more convenient to shoot out of continuity. In 1911, Frank Woods commented upon inconsistent screen direction and its possible production causes:

Woods suggests that: 'It is probable that the trouble is due in some instances to the fact that interior scenes are made first and exteriors cannot always be made to accommodate themselves.'24 Indeed, most changes in screen direction in films of the early teens seem to result from shooting in different locales, then trying to match up the results. Particularly problematic are movements between interior and exterior locales. There are also instances where the different interior sets were not planned to take entrances and exits into account, so that the character repeatedly changes direction between the two settings. Allan Dwan's The Fear (1912, American) shows an instance of a reversed screen direction which must have resulted from out-of-continuity shooting. Throughout the film, we see shots made on a rocky beach; in one such shot, a character exits toward the left front. In the next shot, he should approach from the right side of the frame, but he comes in left.. Since these two spaces are never seen in a single framing, we can assume that Dwan's unit probably filmed in two locations far separated from each other and that no one kept track of this particular movement. As a result, several movements and eyelines between these two spaces are regularly mismatched in the same way.

After 1915, quite a few films continued to violate screen direction in relation to movements, but often only once or twice in an entire feature film - no more than might occur in a film of the thirties or forties. One reason that violations of screen direction occurred at all is that there was no established method for avoiding them until the late teens, when the 'script girl' began to be a regularly assigned position. In the early teens, there seems to have been some attempt to solve the problem by specifying screen direction in the continuity script, extending its function as a blueprint guiding all aspects of production. Two 1913 screenplay advisers give the following instructions: 26

Describe where each character enters or exits when necessary and how and with whom, i.e., tell where he enters or exits - whether from the house, garden, or door - tell how he enters or exits - whether arm-in-arm, frightened, walking, running, mounted, breathlessly, etc.

Describe when and where the characters are to enter the scene, giving the entrance, or the direction. If they are to be in the scene at the beginning of the film, state that they are 'discovered' and give their position.

In each case, the authors seem anxious to get the scenario writer to specify as much as possible about the characters, not only to maintain screen direction, but also to aid in matching on action.

This attempt to make screen direction part of the written plans apparently did not work, however. While some published specimen scenarios of this period mention directions of entrances, most fail to do so. And set designers went on making an occasional set that did not allow for the matching of exit and extrance from one locale to another. The Italian (Reginald Barker, 1915, New York Motion Picture Co.), for example, consistently mismatches movements from the outside of Gallia's house to the inside. Planning, then, did not always ensure proper screen direction.

By the late teens, not only did filmmakers watch for screen direction in shooting, but editors had developed tactics for correcting problems when they arose. Helen Starr's major 1918 Photoplay article on editing discusses screen direction: 30

Note Starr's assumption that the need for maintaining screen direction is 'obvious.' William Hornbeck, an editor who began as an assistant cutter at Keystone in 1917, discusses other ways of covering a problem; in describing how an editor would cut together two shots in which the screen direction was reversed, he recalled: 32

In some cases, editors might even arrange to have shots redone - especially a close shot which would cost little to make and which could cover an error. By the late teens, filmmakers counted on audiences' 'reading' screen direction in specific ways. Starr discussed how an editor could save money on a production; using a recent battle scene as an example, she quotes the editor's description of how the sequence was done: 33

In spite of Starr's casual conclusion, the passage (with its use of the term 'reverse camera shot' to mean basically what it means today) indicates an extensive grasp of the principles of continuity editing. The filmmakers understood and were putting into practice the effects of opposed screen direction and eyelines that the Soviet filmmaker Lev Kuleshov was to study in his famous experiments on editing conducted during the early 1920s. Screen direction, we may assume, had virtually reached the status of a rule by 1917.

If character movement can cue contiguous spaces, so does character glance. Quite early, around 1902, filmmakers began using glances to create optical point-of-view (POV) shots, placing the camera in the spatial position of the character. Then, during the early transitional period, after about 1909, eyeline matches appear in films; in this device, the character glances to a point offscreen in one shot, and a cut reveals the seen space, but not from the spatial position of the character.

The POV shot can show either a portion of the space seen in the establishing shot, or it can show a contiguous space Filmmakers who first employed the POV shot used it in both ways. In the early years, the POV shot was usually indicated not only by position but also by a mask. Grandpa's Reading Glass (1902, AM&B) uses a series of round masks to represent a magnifying glass; the POV shots show details of the larger space of the establishing shot. Similarly, the hero's view through binoculars in The 100-to-One Shot (1906, Vitagraph; see figs 16.38 and 16.39) shows a space already seen in the first shot. But in A Search for Evidence (1903, AM&B), POV shots done through a keyhole mask reveal a contiguous space, a series of hotel rooms behind closed doors.

In rare instances, an unmasked POV shot might be used in these early years. The Runaway Match (1903, AM&B) even contains a POV tracking shot. When the pursuing father's car breaks down, the camera continues tracking back from him along the road; he stands angrily gesticulating toward the camera as it moves to extreme-long shot. A reverse medium-long shot tracks forward following the car containing the eloping daughter and her fianc6, who wave and laugh directly into the camera; the preceding view of the father is revealed as their POV shot. The Runaway Match provides a good example of a device which was later to become a recognizable part of Hollywood's repertory of devices, but which at the time was more likely an isolated experiment.

The unmasked POV shot occurs more regularly from about 1911. Since most camera angles were at this point nearly horizontal, they were not particularly serviceable as POV cues. Nor had filmmakers developed a set of other cues for indicating that the camera occupied the character's place. The glance through a window provided virtually the only such cue, since the window frame within the image placed the character spatially. Thus A Friendly Marriage (1911, Vitagraph) contains a shot of the wife stopping by a church and looking off left front, followed by a view of her husband through the rectory window. In Kirkwood's American Biograph drama, The House of Discord (1913), the heroine stands by the gate of an estate and watches her daughter ride past with a groom; there follows a POV shot of the pair going away down the road. Here the position of the road and the direction of movement of the couple on horseback tell us that the camera has been placed in the heroine's position in the second shot. In Behind the Footlights (1916, Vim Comedies), the hero looks out from behind the curtain of a vaudeville stage and sees his girlfriend in the audience. Here camera angle as well as the placement of orchestra members in the lower part of the frame signals POV. By 1917, most films use POV at least once, usually employing continuity cues of spatial relations from shot to shot to indicate POV, only occasionally including windows and binocular or keyhole masks.

By the late teens, the masked POV shot returned, but not with shapes suggesting binoculars or keyholes. Instead, masking became a conventional means of marking POV shots as such. In *Love and the Law (Edgar Lewis, 1919, Edgar Lewis Productions), one scene contains a lengthy series of POV shots as the hero stands by a parked wagon, turns slowly around, and sees a series of shop signs; there are five shots of him looking in various directions, each followed by a masked shot of a sign. The sophistication of POV usage by the late teens is apparent in this series: each sign is at the precise angle and distance it would be in relation to the character's position, those on his side of the street being closer than ones across the street or further down the street. In accordance with Hollywood's growing use of redundancy, the spatial and masking cues supported each other in indicating POV. In later years, additional POV cues reinforced the principles formulated in the teens.

Capt. Leslie T. Peacocke, 1917

If a single eyeline provides a strong spatial cue, then a second eyeline on the other side of the cut should create an even stronger spatial anchor for the spectator. This principle is commonly used to create the shot/reverse-shot (SRS) schema, one of the most prevalent figures in the classical Hollywood cinema's spatial system. The SRS also depends on screen direction.

As we have seen, the concept of the eyeline match existed by 1913-14. Cuts that change screen direction after a glance were distinctly in the minority. The same is true of the shot/reverseshot pattem. No doubt one can find occasional violations throughout the teens. But this does not indicate the absence of a guideline -filmmakers in the thirties occasionally crossed the line on SRSs as well.

SRS was introduced near the beginning of the transitional period. Early instances of this technique show it already performing its classical function of presenting a conversation situation; balanced pairs of shots form the centerpiece of a scene that contains other contiguous cuts as well. Barry Salt has pointed out" an early example of SRS in Essanay's The Loafer (1911), a film which is generally remarkably advanced in its application of classical principles. The shots he describes come in the middle of a classically constructed sequence which opens with an establishing shot of the hero by a buggy. He has been a drunken loafer, was given a loan, and now is a respectable farmer. After the shot begins, the camera pans right to reframe a stranger approaching the hero to beg for money. After the hero refuses him, the tramp goes out right. A cut reveals a grassy stretch of ground, and the tramp comes in from the front left, turns, and begins to berate the hero. In the next shot, we see the hero's reaction; then he runs out right threateningly. Cut to a shot of the beggar, as he runs out right. There follows a long shot of the field, and both men dash in from the right. The scene continues with the hero running out left after the struggle. Next we see a shot of a farmhouse door, and the hero comes into frame from front right. This relatively extended sequence of nine, combines several movements to contiguous spaces maintaining screen direction, plus a SRS framed three-quarters on each figure, again obeying screen direction. An establishing shot and two reframings give further indication of careful planning along the lines of continuity principles.

SRS became more frequent around 1914, now occurring in some cases between people who are close to each other; a two-shot could easily have been used in these cases, but the director cut in for a pair of closer shots to catch reactions during conversations.

By 1915, SRS had become majority practice, and I found no film from 1916 and 1917 that lacked it. Films of 1915 that use the device range from the most prestigious features (The Cheat [Cecil B. DeMille, Lasky]) to extremely clumsy comedy shorts (Cupid in a Hospital [an L-KO Chaplin imitation]). indicating the widespread adoption of the SRS pattern. Not a single one of the SRS patterns in 1915 ES films violated screen direction."* There are probably films from the late teens which avoid SRS, but certainly the pattern is almost universally accepted by this point Feature films were now using SRS throughout, and not only for distantly separated characters. These examples show characters who are within a few feet of each other and who have previously been seen together in establishing shots.

There is one striking difference between SRS in the teens and SRS as practiced in the 1930s. Sound films often place the camera behind the shoulder of one character when framing the other; shoulders provide one more spatial cue to orient the spectator. Occasional silent films do use shoulders or other portions of the body for such a function, although this remains minority practice until the sound period. Maurice Tourneur's *Victory (1919, Tourneur) provides an early example, and Mantrap (Fleming, 1926, Famous Players-Lasky) uses compositions very similar to those of sound films. Such framings show up not infrequently during the twenties. Thus SRS became one of the most basic devices of the late teens and twenties classical cinema, appearing in most scenes. We may be surprised to find this particular device so common in a cinema in which characters' speech could not be heard, but passages built around dialogue (only partially conveyed through dialogue inter-titles) were an important basis of many silent films. By the late twenties, the handling of conversation situations was schernatized in a way which would barely differ from that of sound films.

The classical cinema's dependence upon POV shots, eyeline matches, and SRS patterns reflects its general orientation toward character psychology. As Part One stressed, most classical narration arises from within the story itself, often by binding our knowledge to shifts in the characters' attention: we notice or concentrate on elements to which the characters' glances direct us. In the construction of contiguous spaces, POV, the eyeline match, and SRS do not work as isolated devices; rather, they operate together within the larger systems of logic, time, and space, guaranteeing that psychological motivation will govern even the mechanics of joining one shot to another. As a result, the system of logic remains dominant.

'crosscutting' is editing which moves between simultaneous events in widely separated locales. 'Parallel editing' differs in that the two events intercut are not simultaneous. Interestingly, crosscutting was seldom used before 1910. In The Great Train Robbery, Porter's narration returns from the robbers' flight to the situations at the telegraph office and dance hall, but he does not alternate shots in these locales. Similarly, in The Kleptomaniac (1905, Edison), Porter first shows the rich woman's actions and then the poor one's, in order to contrast the treatment of the two when they are arrested for stealing, but he does not alternate between them.

The more conventional 'rescue' pattern of crosscutting, involving two persons or groups who eventually meet, occurs at least as early as 1906, in The 100-to-One Shot (Vitagraph). In this film, the hero goes out and wins money on a long shot to aid his fianc6e and her father, who are about to be evicted. As they are being thrown out of their house by the landlord, the following brief series of shots creates suspense:

29 ELS: A street. A car in the distance drives straight forward and out right foreground.

30 LS: Interior of the house, as earlier, but with furniture gone. The landlord enters from the right, and, with the help of two officials, starts to lead the father out right.

31 ELS: A road. The car comes in from the background, drives forward, stops, and the hero gets out and runs out right.

32 As 32: The hero enters from the right (a violation of screen direction), tears up the landlord's paper, and pays him. The villains leave, and the film ends with rejoicing and an embrace.

Another example occurs in Her First Adventure (Wallace McCutcheon, 1908, AM&B), which is generally handled as a conventional chase until toward the end. Then a few shots alternate between pursued and pursuers. From 1909 on, Griffith begins to use the device occasionally and was probably responsible for popularizing it.

Crosscutting did not become widespread immediately, however. By 1912, slightly fewer than half the ES films used any crosscutting. Some of these include chases, as in The Bandit of Tropico (1912, Nestor) and The Grit of the Girl Telegrapher (1912, Kalem). Others simply use crosscutting to show two related events occurring in separate spaces. In The Haunted Rocker (1912, Vitagraph), there is one instance of crosscutting when the disapproving father goes to his club while his daughter's lover visits her secretly at home. The following sequence occurs:

13 LS: The steps outside the house. The father goes out, then the lover goes to the door. 14 MLS: The parlor. The heroine sits in the rocking chair. Her lover enters and they embrace. 15 MLS: Interior of a men's club. The father comes in , has a drink, and leaves. 16 MLS, as 14: The heroine sits on her lover's lap in the rocker 17 MLS: The front gate of the house. The father comes in, drunk. 18 New MLS: The rocker. The lovers stand hurriedly and hide behind a screen. The father enters, sees the moving chair, and is puzzled.

One noticeable trait of this sequence is the considerable compression of time made possible by the crosscutting. At each return to the previous action, a move forward in the narrative has occurred. The crosscutting represents simultaneous events, but also creates large ellipses which are less obvious because of the move away to another line of action. As crosscutting became more common, this ability to shorten plot duration remained one of its most important functions.

Contemporary writers recognized that crosscutting could condense narrative material, as well as create suspense. A 1914 scenario manual referred to the 'cutback' (as crosscutting was known at the time) as being 'employed to accelerate action and maintain suspense. 1311 In 1923 the American Cinematographer described how an editor could reduce an excess of footage to a finished film: 39

Thus by editing the relatively inessential moments of each story line, the omnipresent narration guides the spectator's attention through a string of the most salient actions.

In a sense, this compression through crosscutting carried on the basic approach of the early teens, when short film lengths led to highly condensed presentations of action At that point, summary titles, telegraphic pantomime gestures, and other devices had combined to pack a great deal of action into a short span Now crosscutting could create a similar effect, but in a less obtrusive way.

With the feature film, such extreme condensation of action was not always necessary. Sometimes the opposite problem arose: how to sustain an action through a whole sequence Some filmmakers found crosscutting to be the solution. Crosscutting permitted the action of a single sequence to be drawn out, where showing the actions in separate short scenes might make the film episodic. Cecil B DeMille's feature The Whispering Chorus (1918, Artcraft) is an example This seven-reeler has thirty sequences, eleven of which employ crosscutting between two lines of action which do not come together within any one sequence (as well as two others which juxtapose action in two locales without cutting between them). The story covers a long time span and involves a large number of separate locales and incidents. Without crosscutting, the film would consist of a string of brief scenes; with it, there is less sense of choppiness.

By 1914, most ES films used crosscutting, and after 1915, only a few films avoided it. Once crosscutting had been established, filmmakers continued to add more and more lines of action, the most famous instances being the multiple simultaneous rescues near the ends of The Birth of a Nation and Intolerance Griffith, who was universally assumed at the time to have invented the cutback, was the prime experimenter in this. But apparently even he went too far; a review of his 1918 feature The Great Loue (now lost) comments:40

Crosscutting became standardized as the interweaving of two or three lines of action -seldom more.

The crosscut scene had become a staple of the silent cinema by the late teens and twenties. More often than not, crosscutting provided a simple way of constructing an exciting story without the script writer's having to sustain a single line of action. It seems to have reached its most frequent usage for this purpose in the few years after 1915 By the twenties, script writers had gained more experience at creating situations which could sustain themselves for whole sequences Crosscutting did not disappear, but became a more localized device, occurring mainly in scenes where the narration demanded the juxtaposition of multiple lines of action

The various continuity rules - establishing and re-establishing shots, cut-ins, screen direction, eyelines, SRS, crosscutting - served two overall purposes. On the one hand, they permitted the narrative to proceed in a clearly defined space. On the other hand, they created an omnipresent narration which shifted the audience's vantage point on the action frequently to follow those parts of the scene most salient to the plot.

Two statements from the twenties summarize these purposes succinctly and demonstrate that Hollywood practitioners understood their editing system as fully then as have practitioners ever since. A former editor for Ince wrote in 1922: 'The value of every scene and sequence must be carefully weighed and the man who attempts to do this must most surely be able to prepare and smooth the production for audience consumption. The second statement comes from a lecture given by actor Milton Sills in 1928; he begins by discussing how developing methods limited the length of shots:'3

Thus continuity editing constantly organizes the spectator's attention. In doing so, it acts in concert with other principles of the classical cinema -principles of depth and centering that guide the eye within shots.