Silent Films - What Was the Right Speed? (1980)
By Kevin Brownlow, 1980
The silent film died, commercially, fifty years ago. Since then a swarm of misconceptions have obscured a great deal of fascinating history. Ask people to describe a silent film, and they'll tell you they were 'jerky', like the Chaplins they've seen on television, or they'll talk about 'flicker' and 'bad photography'. The last two charges can often be laid at the door of modern laboratories, for the original prints were generally superior to the black and white they produce today. But the idea that silent films were 'jerky' is less easily dismissed. Shown at the right speed, of course, they move as smoothly as a modern film - but what was the right speed?
Silent films have more mechanical drawbacks than other antiques - you can't even look at them without the right equipment. And the 'right' equipment is generally wrong. 16mm sound projectors are fitted with a switch marked 'silent' and 'sound'. The latter indicates the standard speed for sound films - 24 frames per second. 'Silent' is supposed to do the same for silent films. But it doesn't. It merely indicates the speed for films photographed on clockwork home-movie cameras - 16 or 18 frames per second. The fact that some silents were photographed at this speed is a happy coincidence. But it has given rise to the illusion that all silent films, professional and amateur, were photographed at 16 frames per second. And they weren't. I have interviewed many cameramen who worked on silent pictures, and I have asked them, again and again, about the question of speeds. They always give me the same answer. 'The standard speed was sixteen.' When confronted by evidence that it wasn't, they looked puzzled. But they never shift from their position. And hardly any of them has offered a satisfactory explanation.
To try to sort it out, I have done research in the trade papers of the time. I discovered the controversy popping up throughout the silent era. For the silent film placed responsibility on the projectionist in a way the sound film never did.
This came into sharp relief when I began work with David Gill on the Hollywood series for Thames Television. We were faced with doing the programmes on film, and initial tests with various laboratories proved very disappointing. Laboratories could reproduce neither the tonal range nor the sharpness of the original nitrate prints. What was worse, they could not alter the speed satisfactorily. They could stretch-print the film - by using an optical printer, they could print every third frame twice to give the equivalent of 16 fps - but this only increased the problem. 16 fps was too slow for most silent films, and the stretching tended to give the action a hiccup effect which was most distracting.
After further tests, and long discussions with our supervising editor, Trevor Waite, we decided to abandon the conventional approach and to make the series on videotape. This gave us complications galore, but at least it preserved as much of the photographic quality as the television line system would allow and, even more important, it enabled us to alter the speed without destroying the action. We used a variable speed telecine machine called a Polygon. Because they are virtually obsolete, suitable Polygons in Britain exist only at the BBC - which has both 16mm and 35mm models. Fitted with a 28-sided prism - 'The Flying Ashtray' - the Polygon enables you to transfer film to videotape at any speed from 4 fps to 35 fps.
The Polygon has a deficiency of its own; under about 16 fps any lateral movement is subject to 'image drag'. The Polygon prism actually 'mixes' from frame to frame, so that at slow speeds one's persistence of vision tends to retain the double image, giving the 'drag' effect. This was a drawback we had to accept to a certain degree (no doubt people will blame that on the crudity of the early films, too!). There was a loss of picture quality compared to other telecines, too. In other respects, the Polygon was a godsend. It was particularly fascinating to run the early films, altering the speed until the movement seemed absolutely right. A pattern emerged that seemed related to the year of production, the studio and, of course, the cameraman.
Yet these speeds seldom corresponded with the speeds at which the films were shown in the theatres. It is possible to check these because they were specified on the cue sheets. These cue sheets were issued to the musical directors of the theatres, and contained musical suggestions. The same sheets were available for the projectionists. 'Examination of hundreds of cue sheets for silent films,' said James Card, former curator of George Eastman House, 'has failed to turn up a single one which indicates a film should be projected at 16 frames a second.'
The cue sheets were often quite emphatic. For The Four Horsemen they insisted: 'The correct speed is 12 1/2 minutes per 1,000 ft. - not any slower - and mark off, then make sure and hold it. Speedometer should be used and film should be run registering 86 revolutions per minute.' Speed: 21 1/2 fps. The chart on the next page offers some fascinating comparisons.
The Polygon method is not definite proof of the original camera speed, for it is entirely dependent on the eye of the person in charge. Everyone has a slightly different sense of rhythm; I tended to favour a frame or two faster than David Gill, for instance. And one can check back at our Polygon sheets and see that a film transferred at 22 fps at one session might run at 21 or 23 at another. But while it may not be scientific, the Polygon is the best guide to film speed since projectors abandoned tachometers.
The opportunity for television filmmakers to achieve the correct speed will soon be widely available. The new generation of Ampex vpr tape machines - with variable speed capacity - will give even greater control, for they permit a reduction to the equivalent of 4-5 fps without image drag, and with no loss of definition.
Controversy over speed dogged silent films from the start. Thomas Edison recommended a speed of 46 frames per second - 'anything less will strain the eye.' As historian Gordon Hendricks wrote in his book "The Edison Motion Picture Myth": 'There would seem to be no good reason for it. This rate was far above any rate necessary for gaining the persistence of vision.' H.A.V. Bulleid points out, however, that Edison's decision was a sensible one: 'To obviate flicker from white light projected on a bright surface requires about 48 obscurations per second.' Nevertheless, Edison films did not follow this recommendation for long. Apart from anything else, it used too much film. It also reduced the exposure, and film stock was not fast. But Edison films were photographed much faster than the films of most other companies - although Hendricks found them varying as much as 15 fps in a single day. An Edison film of 1900 will generally project satisfactorily at 24 fps. Edison's rival, the American Mutoscope and Biograph Company, used a camera which weighed 1700 lbs. This camera had a motor, and it turned at a speed of 40 fps. Billy Bitzer operated one. (Curiously, his later films were characterised by a remarkably slow camera speed.)
During the Nickelodeon period, films were projected at whatever speed suited the management. The standard was supposed to be 16 fps. 'I remember running a full 1,000 ft. reel in 12 minutes at the eight o'clock show,' recalled Victor Milner, who later became a leading Hollywood cameraman, 'and in the afternoon I used to project the same reel so slow it took Maurice Costello ages to cross the set. Those were my manager's orders.' Projected at the 'correct' speed of 16 fps, a full 1,000 ft. reel of 35mm film would last 16 1/2 minutes. The Essanay Film Company of Chicago tried to beat wily exhibitors by printing the running time of the films on the posters. The exhibitors retaliated by pasting a strip of paper over the line. Some unscrupulous theatre managers could get through a full reel in six minutes! Ten minutes was acknowledged to be 'more usual'. Yet, even today, on standard 24 fps sound projectors, 1,000 feet takes eleven minutes...
'There is no hard and fast rule that can be laid down governing speed,' said Moving Picture World (9 May 1908, p. 413). 'It may, however, be said that 70 feet per minute is about as fast as a film should be run under any circumstances, with 45 as the limit the other way. Slower than 40 feet would not be safe. In general, the film should be run at the speed that will produce a minimum of flicker, combined with the lifelike, natural motion of the figures... It is as likely as not that the speed should be changed several times in different portions of the same film. With most standard machines, one turn of the crank runs off exactly one foot of film, so that normal speed is about 66 turns of the crank per minute, and by counting turns you know just how fast you are running.'
Projectionists might speed up Edison one- and two-reelers with relative impunity. But Biographs looked ridiculous at anything above the so-called standard 16. By 1913, even that speed was too fast. The chief Biograph director, D. W. Griffith, appeared to be struggling against the limitations of the one- and two-reeler. (The following year, he would embark on his epic The Birth of a Nation in 12 reels.) By slowing the speed of the camera (and therefore the projector) he could squeeze in extra sequences and extend his story. Biograph instructed exhibitors to project their films so that a full 1,000 ft. reel lasted eighteen minutes - 15 fps.
The highly inflammable nitrate film had to move slowly past the searing heat of the arc lamp. On most projectors, the fire shutter would descend and cut off the light if they moved below 40 ft. per minute. Projectionists often ignored the 18 minute-per-thousand rule. One might assume that reports that his actors were zipping across the screen would horrify Griffith, and he would increase the speed of his camera to suit the standard projection speed. Not at all. Some sequences of The Birth of a Nation are so undercranked that they need to be shown at 12 fps. Griffith, and his cameraman Billy Bitzer, continued to crank slower than average on all the major features they made together. And because Griffith's films are the most frequently revived of all American silent films, film societies religiously switch their projectors to 'silent' for Griffith films - and all other silent pictures. The speed is not slow enough for Griffith - and is ruinously slow for other films.
But not even Griffith was consistent. His instructions for Home Sweet Home (1914) recommended 16 minutes for the first reel (16.6 fps), 14-15 minutes for the second (17.8-19 fps), and 13-14 for each of the other reels (19-20.5 fps). 'The last reel, however, should be run slowly from the beginning of the allegorical part to the end' ( Moving Picture World, 20 June 1914 p. 652). 'The projectionist,' said Griffith, 'in a large measure is compelled to redirect the photoplay.'
A 1915 projectionist's handbook declared - in emphatic capitals - 'THERE IS NO SUCH THING AS A SET CAMERA SPEED!' The correct speed of projection, it added, is the speed at which each individual scene was taken - 'which may - and often does - vary wildly.' And it declared: 'One of the highest functions of projection is to watch the screen and regulate the speed of projection to synchronise with the speed of taking.'
The reason that films of this period were sometimes shown too fast was that exhibitors worked to iron-bound schedules. For a 7-reel show, lasting 1 hr. 50 mins., each reel could be run at an average 15.7 mins. per reel (around 17 fps). But if there were 9 reels, the projectionist had to speed up to 12.2 mins. (about 22 fps) in order that the show could end precisely on time. Nevertheless, the fact that silent pictures were invariably shown slightly faster than they were shot was confirmed by historian David Shepard. When for example he worked on the restoration of Nanook of the North (1922), he found that Flaherty had shot the film at 16 fps - 'but the pace and rhythm of the edited feature was completely destroyed by so slow a speed.'
Exhibitors declared in the 1920s that the standard speed was much faster than it used to be, because cameramen varied their rate of cranking so much. The Hollywood cameramen were indignant. Victor Milner stated in the American Cinematographer, in July 1923, that the average camera speed was still 16 fps; sixteen minutes, forty seconds per thousand. 'To achieve smoothness of tempo, projection should also be at this speed.' He pointed out that rushes were shown every day at the studios, on the most modern projectors. The operator had an indicator, and there was another indicator on the director's desk. It was set between 65 and 72 feet per minute. 'If a scene shows the wrong speed, it is retaken.' Already, Milner has weakened his argument. For he has admitted that films, allegedly photographed at 16, were being checked at a speed approaching 19. In the same article, Milner enlisted the support of director Rex Ingram; he declared that he was very careful to achieve perfect speed. Yet the films of Rex Ingram, photographed by John Seitz, registered between 19 fps and 22 fps on the Polygon, and his cue sheets specified from 21-24 fps.
In 1925, director Al Rogell wrote in Director magazine: 'At a recent meeting, the Society of Motion Picture Engineers advocated a universal set speed for projection of 80 feet per minute (21.3 frames per second). As a matter of fact, most theatres show pictures at a speed of 85-90 ft. per minute (22.6 fps - 24 fps) in these days of long shows with 10-11 reels of film and various entertainment acts.'
That same year (1925) the head of a theatre chain in Indiana claimed that 24 fps was now the standard speed. Again the cameramen denied it. Paul Perry, in the American Cinematographer, claimed that he had checked with his colleagues and found the majority cranking at - you've guessed it - 16 fps, 60 feet per minute. The Polygon may not be a definitive scientific instrument, but no film of 1925 we put through had been shot at 16 fps. Yet Perry let slip a hint that cameramen had been asked to turn faster 'If the theatres insist on faster projection, it is only natural that some of the producers request that their film be exposed more rapidly to offset the increased speed of projection.'
Karl Malkames, the son of silent era cameraman Don Malkames, and a cameraman himself, wrote to me recently - 'The practice of cranking from 20 to 24 to compensate for rising theatrical projection speeds was common while my father was at Fox in 1924. Bell and Howell equipment continued to reflect the idea of 16 fps as "normal". The shutter speed plate on my later model Bell and Howell, circa 1930, is still calibrated for 16 fps.'
The Bell and Howell 2709 camera became the standard studio model of the 1920s. Two turns of the handle sent one foot of film through the gate - one foot of film contained sixteen frames of picture, so two turns per second equalled sixteen frames, the 'standard' speed. Cameramen prided themselves on their even rate of cranking. To achieve it, you said 'One hundred and one' to yourself, over and over, until the speed came naturally. One cameraman blamed the 'Anvil Chorus' from II Trovatore. While the speed was undoubtedly even, how could they know precisely what their speed was? Could they honestly say they were cranking at 16 and not 18 frames per second? For there was no speed indicator on the Bell and Howell.
Kemp Niver has interviewed a number of silent-era cameramen for the American Society of Cinematographers. 'To a man, they said that unless there was some specific reason... they tried to maintain 16 fps. No cameraman could make a statement that he cranked at a specific speed and prove it unless he had a stop watch at the start and end of a given number of feet of film. Even then he couldn't guarantee that he wasn't undercranking or over-cranking as the tension on the take-up magazine increased.'
A motor was available for the Bell and Howell, with a speed indicator. Predictably, 16 fps is marked NORMAL. The top speed is 22 fps. In production stills, this motor appears more and more often towards the end of the 20s. In the old days, they said they preferred hand cranking. So why the motor? Perhaps because of the increased demands of the mobile camera - now far more common - which needed more manipulation than the static camera. And perhaps because a standard speed had at last arrived. In October 1927 The Jazz Singer had been premiered, and theatres were being wired for sound. The standard speed of sound films was 24 fps. It is interesting to discover how that speed was arrived at.
According to Stanley Watkins, head engineer for Western Electric, he and his team checked with the Warner Theatre for the average speed of projection in 1926. They were told 'between eighty and ninety feet per minute' in the big theatres - between 20 and 24 fps - and around 26 fps in the smaller theatres. They settled on 90 feet a minute (24 fps) as a reasonable compromise for the Vitaphone process. The other sound systems began at slower speeds (Fox-Case's first tests were shot at 21 fps), but they, too, adopted 24 fps as standard in November 1926.
If Hollywood cameramen were still working rigidly at 16 fps, their work would have looked ludicrous in the public theatres. The fact that they were cranking the films slower than they were projected is borne out by The Jazz Singer. Al Jolson walks to the stage at a slightly accelerated pace, and when the Vitaphone section begins, he is filmed at 24 fps. Filmgoers often remarked that the early talkies seemed leaden-footed.
While filming Annapolis in 1928, cameraman Arthur Miller received a wire from the studio to crank at 24 fps. He did so, and everyone complained that the speed slowed everything down too much - it was particularly noticeable in a dress parade scene of midshipmen.
Walter Kerr, in his brilliant book 'The Silent Clowns,' puts forward his theory: silent films were photographed at 16 or 18 frames a second, but projected at a rate closer to sound speed. 'The result was not only faster than life, it was cleaner, less effortful, more dynamic.' Kerr illustrates his theory with a photograph of the instructions printed on the leader for the 1922 Down to the Sea in Ships: 'Operator: please run eleven minutes for 1,000 ft.' - or 24 fps, the speed of sound. I have projected the film, and found two or three sections too fast at that speed (although the remainder is satisfactory). But that was how audiences saw it at the time. Other, later films, such as The Winning of Barbara Worth, were shot at a speed so close to 24 fps that they are wrecked by being shown at 16 fps. I have even seen an occasional silent, such as The Blood Ship (1928) apparently designed to be projected at 26 fps, since sound speed is too slow for them. (Poverty Row producers like Columbia, who made The Blood Ship, ordered cameras to be cranked faster to fill their reels more economically!)
Talking about comedies, Walter Kerr writes: 'Silent films chose, by control of the camera and through instructions to projectionists, to move at an unreal, stylised, in effect fantasised rate.' And he quotes as examples the last two silent Chaplins - both films designed to be shown at sound speed, but photographed at silent speed (whatever that was!). 'The least glance at Modern Times reveals instantly that all of Chaplin's work in the film... has been filmed at a rate that puts springs on his heels and makes unleashed jack-knives of his elbows. This is how the films looked when they were projected as their creators intended.'
One can only have sympathy for those who programme silent films. But sympathy evaporates as soon as one has to endure slow projection. It's bad enough to be deprived of the sound of the symphony orchestra (which accompanied all first-run films in the big theatres), but to be forced to watch the films in dead silence at an equally deadly pace is too much to ask of anyone. William Wellman's spectacular war film Wings (1927) moves at an exhilarating pace when projected at the speed at which it was shown originally - 24 fps - but it drags miserably when shown at 16. At one of its last major showings in Britain, a few years ago, the audience emerged complaining of its slowness. Their complaints were not aimed at the projectionist, for they thought it was an inherent fault of all silent films. Yet it had been shown at 16 fps - and had lasted nearly an hour longer than in 1927. That attitude shows no respect towards the films of the past. It does them a grave disservice.
'Ideally,' says David Gill, 'projectors should have variable speeds from at least 14 to 24 fps. There was no standard for the normal camera speed in the silent days, and for commercial reasons, the recommended projection speeds do not necessarily match the camera speed. We could recommend that most silents be shown at 24 - but that means that silent film actors will always move in crisp, sharp movements - never smooth and languorous ones. This is fine for actors in comedies, war films and some Westerns. But it's sad for the others - especially the Great Lovers!'
Not only was there no standard, but the alert projectionist was expected to dynamically vary the speed depending on the nature of the scene being shown!
fka. Bullwinkle J. Moose