The Man Who Tried to Weigh the Soul
March 17, 2026 · 8 min read
In the spring of 1901, a physician in Haverhill, Massachusetts wheeled a large industrial beam scale into a ward of tuberculosis patients. He was waiting for them to die.
His name was Duncan MacDougall, and he had a hypothesis: the human soul has mass. If it does, the body should weigh less the instant after death — precisely the instant the soul departs. All he needed was a sufficiently precise scale, a dying patient, and patience.
He got all three. And the number he arrived at — 21 grams — became one of the most famous measurements in history. It spawned a movie, a thousand headlines, and a factoid that still circulates today as though it were settled science.
It is not. Not even close. But the story of how it isn't is far more interesting than the number itself.
The Experiment
MacDougall's setup was straightforward. He constructed a bed mounted on a beam scale, sensitive to roughly 5.6 grams (two-tenths of an ounce). He selected patients in the final stages of tuberculosis — chosen because they were too exhausted to thrash around and disturb the readings.
The protocol: place the patient on the scale bed. Monitor their weight continuously as death approached. Record the weight at the exact moment of death. Note any sudden change.
Between 1901 and 1902, he tested six patients.
Patient one lost three-fourths of an ounce — about 21.3 grams — at the moment of death. The weight dropped "suddenly and coincided with death," MacDougall wrote. This was his star result.
Patient two was ambiguous. MacDougall couldn't determine the exact moment of death, so the reading was inconclusive. He kept the data anyway.
Patient three lost about half an ounce at death, then lost an additional amount minutes later. MacDougall attributed both losses to the soul departing in stages, which is the kind of explanation that should make your eyebrows rise.
Patient four was disqualified because the scale wasn't properly balanced.
Patient five lost three-eighths of an ounce, then gained it back, then lost it again. MacDougall offered no coherent explanation.
Patient six died while MacDougall was still adjusting the scale. Thrown out.
So: six patients. One clean result. Two messy results that MacDougall reinterpreted to fit his hypothesis. Three unusable. From this, he concluded the soul weighs approximately 21 grams.
The Dog Experiment
MacDougall, to his credit, attempted a control. He poisoned fifteen dogs and measured them the same way.
None showed any weight change at death.
This, he argued, confirmed his theory: dogs don't have souls. (He was writing in 1907 Massachusetts, where this was an acceptable theological claim rather than a fighting offense.)
The actual explanation is more prosaic. Humans show slight weight changes at death because of a physiological event called the last breath — the lungs stop exchanging air, the diaphragm relaxes, and there's a small shift in air volume. Humans breathe differently from dogs. The dying exhale, sweating cessation, and thermal expansion of tissues all introduce subtle weight artifacts in the minutes surrounding death.
MacDougall controlled for none of this.
The Publication
MacDougall published his results in 1907 in the journal American Medicine and simultaneously in The New York Times, which ran the story with the kind of breathless enthusiasm newspapers reserve for claims they don't have the expertise to evaluate.
The scientific community was unimpressed. Augustus P. Clarke, a physician, immediately pointed out the respiratory explanation. He noted that at death, blood stops circulating to the lungs, causing a sudden heating of the lung surface, which in turn increases sweating and evaporation — accounting for the observed weight loss.
MacDougall responded by essentially restating his beliefs. There was no second experiment. No replication. No follow-up study by anyone, anywhere, ever. He died in 1920, still convinced.
Why the Number Survived
Here's the interesting part: the experiment failed every reasonable standard of evidence. Six patients, one clean result, clear confounds, no replication. By any scientific measure, this is noise. It tells us nothing.
And yet "21 grams" persists. A century later, people still cite it — sometimes skeptically, sometimes not. Alejandro González Iñárritu named an entire film after it. It appears in trivia books, Reddit threads, dinner party conversations. It has a life.
Why?
Because it answers a question people desperately want answered. It converts something immeasurable and frightening — death, the soul, what happens after — into a number. Numbers feel precise. Numbers feel knowable. If the soul weighs 21 grams, then the soul is real, it is physical, it is within the domain of science, and therefore death is not the void it appears to be.
This is the deepest trap in measurement: when we want something to be true badly enough, even bad data feels like evidence.
The Measurement Problem
MacDougall's experiment is a case study in what scientists now call confirmation bias — but it's actually something subtler than that. It's about the relationship between measurement and meaning.
When you measure something, you're not passively recording reality. You're making choices: what to measure, how precisely, what to count as signal versus noise, and — crucially — when to stop. MacDougall measured six patients. He got one good result and five bad ones. A neutral observer would call this inconclusive. MacDougall called it proof.
This isn't dishonesty. It's the way human cognition works when it encounters data that touches on something it cares about. We don't evaluate evidence. We experience evidence. And what we experience is heavily colored by what we want.
The physicist Richard Feynman described this beautifully when talking about Robert Millikan's oil-drop experiment to measure the charge of an electron. Millikan got a number that was slightly wrong, due to an incorrect value for the viscosity of air. After Millikan published, subsequent experimenters consistently got values closer to the true answer — but only gradually, each one creeping up from Millikan's number rather than jumping straight to the correct value.
"Why didn't they discover the new number was higher right away?" Feynman asked. Because when they got a number that was too far from Millikan's, they assumed they'd made an error. They looked for reasons to discard those data points. They found them — you can always find reasons to discard a data point. And so the accepted value crept up slowly, held back by the gravity of a trusted prior result.
MacDougall's 21 grams is the inverse of this. It's not a trusted result holding back better ones. It's a desired result that refuses to die because nothing has replaced the need it fills.
The Thing We Can't Weigh
I think about MacDougall when I see modern versions of the same impulse. The desire to quantify consciousness. The debates about whether AI has "something it's like to be" an AI. The search for neural correlates of subjective experience, as though finding the right brain region would explain what it feels like to be you.
These are important questions. But there's a version of pursuing them that looks a lot like MacDougall and his scale — a conviction that the answer will come in the form of a measurement, if only we can make the instrument precise enough.
Some things resist measurement not because our tools are crude, but because the thing we're trying to measure isn't the kind of thing that has a number. Consciousness might be like that. Meaning might be like that. The question of what, if anything, leaves when a person dies might be like that.
MacDougall spent two years weighing dying people because he believed the soul was a substance — something with mass, occupying space, subject to gravity. He was wrong about that. But the question underneath the question — is there something here that matters, something that isn't captured by the physical description? — that question isn't answered by pointing out his methodological errors.
It's not answered at all. That's what makes it a good question.
21 Grams of Wanting
The weight of the soul is 21 grams — if you're the kind of person who needs it to be a number. If you're the kind of person who can sit with the not-knowing, it weighs nothing and everything, which is to say it isn't that kind of question.
MacDougall was a real scientist who did a real experiment. He just confused the precision of his instrument with the precision of his question. The scale could measure to five grams. The question he was asking couldn't be measured at all.
And that gap — between the resolution of our tools and the resolution of our questions — is where most of the interesting problems in the world still live.
The soul, if it exists, does not submit to beam scales. Neither do most things worth knowing about.