One sometimes hears that statistical data are far-sighted, that if average numbers can show us something it is only a big average picture, while the whole interest about doing film history is in looking at individual films. I totally agree with the latter point and disagree with the former. Yes, the average shot length is a statistical index, but this does not mean that it binds us to static data or average films. I tried it recently to examine a picture as off-beat as Griffith’s Intolerance and it helped me to learn a few things about its editing style which I don’t think I would have been able to find out by any other means.

In principle, what we know about the metric structure of Intolerance from having watched this film and having read about it is enough to get a good sense of what Griffith was trying to achieve. We know that Griffith cross-cuts four stories from four ages; that towards the end of the film the back-and-forth between its stories tends to quicken its pace; and that this quickening is reinforced by the fact that individual shots tend to become shorter and shorter. We may interpret these data differently, of course, Eisenstein in his way, Gunning in his, and Miriam Hansen in hers, but what I am saying is that for a sound historical interpretation of Intolerance we do not need to know more than this.

It was not so much an interpretative need that urged me to push this knowledge further but rather a curiosity about film metrics as such, about its limits of relevance. Let’s see what happens if we gather the shot length data about Intolerance as a whole, and about each of its stories separately, and assess their fluctuations within the duration of the film – will this result in a disorderly (and therefore irrelevant) array of data or will it show a set of regularities, a pattern which confirms, contradicts or corrects what we’ve always assumed about the editing of this film?

One question the timing and counting allows to address is whether or not the average shot length changes depending on the type of the story Griffith deals with and on the epoch in which it is set, in other words, if there is a correlation between cutting rates and subject matter. If there is none, the average shot length within each story will be the same as it is throughout the film, but if there is, it may be worth asking which story is the fastest – the Modern, the Judean, the French or the Babylonian?

The answer is: while the average shot length of Intolerance as a whole is 5.9 seconds, the average shot length of each storyhas this distribution:

  • 1st place: the French story (4.9 seconds)
  • 2nd place: the Modern story (5.6 seconds)
  • 3rd place: the Babylonian story (6.5 seconds)
  • 4th place: the Judean story (6.7 seconds)

I do not think many people will be surprised to find out that the Biblical story is the slowest; but that the Modern story loses to the French one sound counter-intuitive. I do not think it is our intuition that fails us here, but rather, the method itself, for each time we strike an average we level down the extremes. The reason why the average speed of the Modern story is lower than that of the French one is not because it were poorer in short shots – there are enough short shots in both – but because it is richer in long ones: the longest French shot runs for 32 seconds, the longest Modern one, for 53. It is due to this contrast between the fast and the slow that the Modern story feels more dynamic than its average shot length tends to show.

Yes, average numbers are not dynamic, but this does not put statistics out of court. If you tell your laptop to represent your data not as a number, but as a graph, the result will give us exactly the dynamics that the numbers missed. There exist two types of graphs, or trendlines, which statisticians use for their ends, and which we can use for ours. One is called linear, another, polynomial. The linear trendline is a straight line that shows a general tendency, and the polynomial line is a curve that shows a dynamics.


The straight dotted line in this graph shows that as a general tendency the cutting rate of Intolerance climbs during the film; and what the two-humped polynomial curve says is that this tendency is not steady, that the film starts slowly, has two waves of activity, a minor and a major one, and slows down at the end – a dynamics that complies with an old theory according to which a well-crafted drama (or a story, or a film) must start calmly, have two climaxes and resolve in a quieter coda. I find this picture useful, but not indispensable, for people who know Intolerance well could tell without looking that there must be something like an upsurge in film’s tempo around the place when the troops attach the strikers in the Modern story, and another one when the Persian troops attack Babylon; and of course, there is a peaceful apotheosis which is responsible for the slowdown in the end.





A more interesting picture will emerge if we look at the metric profiles of each of the four stories taken separately. While three of them comply with the film’s general tendency to pick up the pace, the Judean story (the slowest of the four) slows down as it follows Christ from the wedding at Cana to the cross. My guess is that this anomaly may be due to an interference of a generic norm to which the Christ story must conform. It may take us to gauge the shot lengths of some other Passion plays (including perhaps the recent Mel Gibson version) to demonstrate this, but it may well turn out that all Passion plays tend to slow down their pace towards the end to be able to relate the last events of Christ’s life in their painful details.

Also: there is an interesting similarity between the metric profiles of the Modern and Babylonian stories: both go up and down, then again up and down. Does this pattern reflect some kind of general rule of dramatic rhythm or perhaps this is Griffith’s patent way of shaping the narrative flow of his films? – Again, the future may show, for to answer this we’ll need to examine more metric data from more Griffith movies, and not only his. But if there is a regularity to discover, it’s worth a try.

And another thing. Note that the curve of the French story does not dive towards the end as the other three stories do, in other words, that this story never slows up. This is not hard to explain knowing that the French story ends in medias res, as it were: Griffith quits this story before the massacre is over. A trickier question might be what makes him do so and it is here I think that metrics can help us account for the subject matter instead of the other way round.

I do not think anyone will disagree if I say that leaving it off in the heat of a battle in not Griffith’s normal way of ending a story, so little so that his biographer Richard Schickel has tried to explain this anomaly by a loss of a miss on Griffith’s part: “As for the French story, it has a truncated feeling about it, as if, perhaps, Griffith shot more of it than survived the final cut.”* It seems more likely, however, that Griffith intentionally sacrificed a neat narrative closure of the French story to maintain the metric flow of Intolerance as whole (see Fig. 1). The French story ends only some 15 minutes before the end of the film as a whole, and if Griffith decided to close it off with his usual slow-down it would against the general climax he was building. To use Ludwig Wittgenstein’s favorite metaphor, Intolerance is a motor-car with a four-stroke engine, and no good engineer would allow one of its cylinders to slow down as the car is gathering speed.


It is this unique feature of the narrative style of Intolerance, the teamwork of its fours stories, which the elegant sinuous line on fig. 5 tells us about. Remember, what makes Intolerance different from other multistory narratives is that Griffith keeps jumping back and forth between its stories. The data which this diagram sums up are not shot lengths as in previous cases, but the length of the story chunks which Griffith cuts between as the film evolves. The line heaves where the cuts between the stories become more frequent, and where they get less frequent the line sinks. See how clever Griffith’s editing is. He begins with relatively brief story chunks in order to bring it home as early as possible that there is a connection between the four epochs. This done, Griffith affords to linger on each of stories longer, to give it time for a proper exposition (primarily on the Modern and Babylonian ones, for these two are by far the longest), which is why the line ebbs till around the middle of the film. But the higher the tension within each of the stories, the more Griffith switches between them. This, again, is followed by a slower coda. It is only when we see his editing at a glance – as a graph, that is – we can see why Intolerance is a masterpiece of timing and temporal composition.

*D.W. Griffith: An American Life, New York: Simon & Schuster, 1984, p. 314.