Sixteen Frames Or Eighteen. Forty Centimetres Hangs On A Dial Nobody Read.
We surveyed the sandbar with a total station, borrowed the camera, had its lens measured on a bench, and gave the lot to three photogrammetrists working blind. All three returned a height that depends on a setting nobody wrote down.
A standing series in which this newspaper investigates a widely held claim and reports that it could not substantiate it.
BLUFF CREEK, Calif. — At sixteen frames per second, the subject in the 1967 film is about 1.95 metres tall. At eighteen frames per second, the same subject, in the same film, on the same sandbar, is about 2.35.
Forty centimetres. That is the whole dispute, and it is not a dispute about the film. It is a dispute about the position of a dial on the side of a camera that nobody looked at on the day.
This newspaper set out to settle the height. We have failed. This is the account of the failure.
What we bought
A total-station survey of the sandbar: two days on the ground, 1,140 points, tied to two features that appear in the 1967 footage and are still in place — a bedrock outcrop on the far bank and a stump. Closure error across the network, 11 centimetres.
The camera, borrowed for eleven days from the collector who has owned it since 1993. We had the lens measured on an optical bench at a metrology laboratory rather than trusting the barrel: focal length 25.1mm, against the 25 stamped on it. That 0.4 per cent matters more than it sounds.
Then we gave the survey, the lens figures and a scan of the footage to three photogrammetrists, engaged separately, each told nothing about the others and nothing about what we hoped for.
Three answers, which turned out to be one answer
All three came back with a line rather than a number.
Subject height as a near-linear function of the assumed frame rate: about 1.75 metres at 15 frames per second, about 2.55 at 19. Roughly twenty centimetres of subject for every additional frame per second. The three functions agree with each other to within three centimetres across the entire range.
That agreement is a real result and we do not want it lost. The geometry is solved. The ground is measured, the lens is known, the subject's position in each frame is fixed to a few centimetres.
Everything in the calculation is measured except the one input that was never written down.
The dial
The camera has detented speed positions. Which one it was on that afternoon was not noted, by anyone, at the time or in the week after.
When it reached us the dial was set at 24. This means nothing. It has been handled by dozens of people across six decades, at least four of whom have said in print that they turned it.
The cameraman gave three different figures over forty-three years. Shortly after the event he said 18. In a filmed interview in the early nineties he said 16 and added, unprompted, that he was not certain. In 2010 he said he had always run it at 24 and that the film had been transferred wrongly ever since.
The third answer is the interesting one, because it is incompatible with the footage, including his own.
The horse
There is one piece of internal evidence and it is a horse. During the pan-out, a horse walks across 41 frames.
A walking horse has a cadence, and cadence is a clock. We asked all three analysts whether that clock could separate 16 from 18.
All three said no, in almost the same words. Forty-one frames, with the animal partly obscured in nine of them and the camera moving through all of them, gives a confidence interval on the implied speed that spans both candidates and reaches most of the way to 24.
"It is a fourteen per cent difference in speed, measured over two and a half seconds of a horse," one wrote. "If you want that to settle it, you are not asking the horse. You are telling it."
What we are printing instead
Not a height.
We are publishing the point cloud, the lens measurement, the three reports entire, and the function. Anyone may check our arithmetic, and we would prefer it if somebody did.
And we would like one clause added to the field, permanently. Anybody who publishes a height from this film should state which dial position they are assuming, in the sentence where the figure appears. Not on page nine. Beside the number.
Every height ever published from this footage is, underneath, a statement about a setting. Sixty years of argument is two groups of people who preferred different values for it and did not say so out loud.
Our correspondent went to the sandbar wanting a number. He came back with an axis.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We stopped arguing about the height and asked what a height figure is made of. We commissioned a total-station survey of the ground, had the camera's lens measured on a bench rather than trusting the number stamped on it, and handed the identical package to three photogrammetrists who did not know of one another and were told nothing about what we wanted. All three reports are published entire, including the fact that all three refused to give us a single number.
Who we spoke to
- Petra Halvorsen-Achebe, Licensed land surveyor. Commissioned by this newspaper; two days on the bar, 1,140 points, control tied to two 1967 datum features May 2026 Her closure error across the network is 11 centimetres. The raw point cloud is published.
- The camera, In one private collection since 1993; lent to this newspaper for eleven days. Focal length measured on an optical bench at a metrology laboratory; governor speeds logged at 4°C, 18°C and 31°C June 2026 Measured focal length 25.1mm, against 25 stamped on the barrel. The speed dial arrived set at 24, which means nothing: at least four people have said in print that they turned it.
- Teodora Vashchenko-Muir, and two other photogrammetrists, Engaged separately; none knew of the others. Each given the identical survey, lens figures and film scan, with no indication of what we hoped for June–July 2026 All three reports published entire. Their height functions agree with one another to within 3 centimetres across the whole frame-rate range.
- Dr. Ottoline Verrall-Nakamura, Image analyst; published a height figure for this film in 2004. Sent the full draft and all three reports; replied in writing at length July 2026 Printed unedited. Her third point is a concession and her second is a request we could not fulfil.
- The camera's owner, Private collector; asked not to be named. Interviewed in person when the camera was collected and again when it was returned; asked in writing about a related reel June and July 2026 Lent the camera on one condition: that we say in print that lending it does not make him a believer in anything.
What we could not confirm
- The frame rate. This is the entire story. No contemporaneous note records the dial position, the camera was never speed-tested at the time, and the film stock's edge markings do not encode running speed.
- That the camera we measured is the camera that was used. Its chain of custody is documented from 1972 and asserted before that. We have printed the gap rather than pretending to a chain.
- The horse. Forty-one frames of a walking animal, partly obscured in nine of them, cannot discriminate 16 from 18 at any confidence we would print. A longer sequence might. We could not obtain one.
How Others Covered This
The same events, as reported elsewhere on the same day. We list what each outlet had that we did not, as well as what we had that they did not — including where we come off worse. Why we print this.
-
The Meridian TelegraphNewspaper Survey Puts The Figure At 2.35 Metres
Took the top of our published range and reported it as our conclusion.
Had that we did not
The survey, the cost, and the fact that we paid for it ourselves.
Left out
The four words 'at eighteen frames per second', which appear beside that figure in every document we published.
-
ApertureThe Length Of A Stride
An essay on why a measurement feels like a verdict, and on what people want from a number.
Had that we did not
The best paragraph written about this film in a decade, on the difference between seeing and settling.
Left out
That the frame rate was never recorded — which is the essay's real subject, and it walks past it twice.
-
They Buriedthis newspaperSixteen Frames Or Eighteen. Forty Centimetres Hangs On A Dial Nobody Read.
Commissioned a survey and three blind photogrammetric analyses, failed to establish a height, and printed the failure.
Had that we did not
The point cloud, the lens measurement, all three reports entire, and the function they agree on.
Left out
We spent four months and a five-figure sum to establish that a number rests on an unstated assumption, which is what numbers do. Dr. Verrall-Nakamura says so in her reply and this newspaper has no answer to it. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Dr. Ottoline Verrall-Nakamura, image analyst on 9 July 2026, in writing, with the full draft, the point cloud and all three photogrammetric reports attached and no deadline. Replied 21 July 2026. Printed in full and unedited.
The survey is good work and I will be using it, and I want that said before the disagreement, because the disagreement is going to take longer.
You write that the frame rate is a free parameter. It is not free. It is bounded, and the bounds are tighter than your piece allows.
The camera has detented speeds. Twenty-four is excluded by your own survey: at 24 the subject's stride length implies a walking speed that does not match her measured progress across your sandbar between frames 352 and 364. So we are at 16 or 18, which you say, and then you stop, as though a two-way choice were the same as an open interval. It is not. It is a two-way choice with evidence on it.
First, the reel. There is footage from the two preceding days on the same reel. In 2004 I estimated frame rates for three of those sequences from water motion and from a man mounting a horse, and all three sat closer to 18 than to 16. A person who has loaded a reel and set a dial does not usually get up the next morning and change it.
Second, your horse. You report that 41 frames cannot discriminate, and that is true, and it is also a choice of test. There are roughly 190 frames of the same horse walking elsewhere on that reel. I have asked for access to that material for eleven years and been refused, most recently in 2023, by the same collector who lent you a camera for eleven days. You did not ask him for the reel. I would like that sentence in your piece, please. It is the most useful thing I can give you and it costs you nothing but embarrassment.
Third, and this one is mine. You are right that I did not state the assumption in my abstract. I stated it in the methods, on page nine, in one sentence, and the figure 2.28 metres has been quoted several hundred times since without it. That is my fault and not yours, and if your piece changes one thing I would like it to be that.
What I object to is the framing. You have discovered that a measurement rests on an assumption, which is what a measurement is, and you have printed it as a sixty-year scandal. The effect of a piece like this is not that people become careful. The effect is that a stranger at a conference tells a graduate student the height was made up. It was not made up. It was computed, from an assumption, which was stated on page nine, by someone who would have been delighted to be asked.
Published unedited under our right-of-reply guarantee.
How was this story?
We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.
Readers' Letters 0
Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.