The number 21 does not appear in Duncan MacDougall’s paper. What he reported, in April 1907, was three-quarters of an ounce: one measurement, from one patient, out of six. The gram figure is a conversion, spread through newspaper summaries and fixed in place when a 2003 film used it as its title.
The argument MacDougall joined was already five thousand years old. People had been trying to weigh the soul long before he arrived.
The scales were already there
The oldest surviving image of the practice comes from Spell 125 of the Egyptian Book of the Dead, first attested around 1475 BCE under Hatshepsut and Thutmose III. The Papyrus of Ani, painted around 1250 BCE and now in the British Museum, shows the scene in detail: Anubis kneels beside a balance scale in the Hall of Two Truths. On one pan sits the Ib, the heart, which in Egyptian theology carried the moral record of a life. On the other sits the feather of Ma’at, the principle of truth and cosmic order. Thoth stands with his reed pen, ready to record the result. Ammit waits below the scale, a hybrid of crocodile, lion, and hippopotamus. If the heart is heavier than the feather, Ammit devours it, and the dead person ceases to exist in any form.
Egyptian virtue was measured in lightness: the pure heart weighed less than Ma’at’s feather. The corrupt heart, burdened by its own record, sank.
The Papyrus of Ani, the finest surviving example of the Egyptian Book of the Dead, is 78 feet long and was painted around 1250 BCE for a royal scribe named Ani. It is kept at the British Museum. The full weighing-of-the-heart scene is depicted across a full scroll section.
In Book XXII of the Iliad, Zeus lifts golden scales and places on them the fates of Achilles and Hector. Hector’s lot sinks. The scene makes no moral judgment: the fates sink or rise, and that determines who dies. The weight is not earned through virtue or vice. Aeschylus wrote a play called Psychostasia in which the mothers of Achilles and Memnon appeal to their divine parents as the scales determine which son lives. The play survives only through Plutarch’s description. In Greek hands, measuring a soul determined its claim on existence, not its moral standing.
The New Testament says nothing about weighing souls at judgment. The tradition appears first in the Testament of Abraham, a Jewish-Christian pseudepigraphic text written in Roman Egypt sometime in the first or second century CE, where an angel named Dokiel holds the scales. The proximity to Egypt matters: the text emerged from a culture that had lived with psychostasia imagery for over a millennium. By the Romanesque period, the image had traveled north into stone. The tympanum at Autun Cathedral, carved by Gislebertus around 1130, shows an angel and a devil contesting the scales. Rogier van der Weyden’s Beaune Altarpiece (commissioned 1443, now at the Hospices de Beaune, Burgundy) shows the Archangel Michael in a red cope holding golden scales over the assembled dead, one pan tipping toward damnation.
Surah 7:8-9 states the Islamic version plainly: “The weighing on that Day will be just. As for those whose scale will be heavy with good deeds, only they will be successful.” Surah 21:47 extends this to the weight of a mustard seed. The Mizan, the cosmic scales, operates as a principle of divine justice with no named angelic weigher. The metaphor runs opposite to Egypt’s: Islamic virtue is heaviness. Good deeds accumulate as mass, and the heavy scale wins. Where Egypt demanded a heart lighter than a feather, the Quran counts every gram of good you have managed to accumulate.
The connection between soul and breath runs through every language that has tried to name it. Greek psyche comes from psykhein, to breathe, and Latin anima shares a root with anemos, wind. Spiritus derives from spirare, to breathe. Sanskrit prana names breath as life force. The Hebrew word God breathes into Adam’s nostrils in Genesis 2:7 is neshamah, breath of life. The Roman ritual of conclamatio had family members press their lips to a dying person’s mouth to catch the last breath. In this conception, breath and soul are the same thing. Removing the breath removes the person.
The weighing of souls was already ancient. MacDougall gave it a hospital bed and a balance beam.
The physician from Haverhill
Duncan MacDougall was born in Glasgow in 1866 and trained at Boston University School of Medicine, graduating in 1893 as class president and class orator. He settled in Haverhill, Massachusetts, where he practiced general medicine and surgery on Main Street.
He was not a Spiritualist. The Haverhill Evening Gazette described him as “hard-headed and practical” with “a disinclination to believe in Spiritualism or psychic phenomena.” His reasoning was strictly materialist: if the soul has substance, it has mass; if it has mass, a scale can detect it. He designed a platform bed mounted on a large balance scale, placed it in a room with dying patients, and began observations around 1901.
The world he published into was one where this project could be taken seriously by serious people. James H. Hyslop, a former Columbia philosophy professor, had reorganized the American Society for Psychical Research in 1904 and launched the Journal of the American Society for Psychical Research in January 1907. Membership grew from 170 to 677 in eighteen months. The ASPR attracted scientists and physicians who believed that consciousness and survival deserved experimental investigation rather than dismissal. Spiritualism had shed many of its parlor-medium trappings and was presenting itself in the register of laboratory research.
The New York Times broke the story on March 11, 1907, weeks before the journals published, with the headline “Soul Has Weight, Physician Thinks.” MacDougall had a known habit of alerting local newspapers to his accomplishments. The formal paper appeared in American Medicine in April 1907 under the title “Hypothesis Concerning Soul Substance Together with Experimental Evidence of the Existence of Such Substance,” and simultaneously in the Journal of the American Society for Psychical Research.
Six patients and what they showed
MacDougall’s scale was sensitive to two-tenths of an ounce (approximately 5.6 grams). He selected patients with tuberculosis or other wasting diseases because they would lie still near death rather than thrashing on the apparatus. He observed six deaths.
Case 1 he called his opportunity. A man with tuberculosis, observed for three hours and forty minutes. At the moment of death, the balance beam dropped and struck its lower limit with what MacDougall described as “an audible stroke.” The loss: three-quarters of an ounce, 21.262 grams. He had already calculated the rate of weight loss from evaporation before death at roughly one ounce per hour. The sudden ¾-ounce drop at the exact instant of death was far faster than any evaporative process could account for, he argued. He lay on the scale himself and exhaled forcibly, registering nothing. This was the only case he considered to have occurred under ideal conditions.
Case 2 presented a ½-ounce drop, but it came fifteen minutes after apparent death. MacDougall admitted in the paper: “We had great doubt…to say just what moment he died.”
Case 3 produced a ½-ounce drop at death and then a full additional ounce minutes later. The second drop coincided with a colleague’s stethoscope making contact with the apparatus.
Case 4 was discarded. The scale was miscalibrated, and MacDougall noted “a good deal of interference by people opposed to our work.”
Case 5 showed a 3/8-ounce drop at death, followed by fifteen minutes during which the scale refused to return to equilibrium, which MacDougall could not explain.
Case 6 was discarded. The patient died within five minutes of being placed on the bed, while the beam was still being adjusted. A 1.5-ounce loss was recorded but attributed to the adjustment.
Final tally: two discarded outright, one clean positive, three ambiguous. MacDougall presented Case 1 as the evidence and the rest as corroboration. Then he wrote: “I am well aware that these few experiments do not prove the matter any more than a few swallows make a summer.”
The dog experiments followed. Fifteen dogs, weighing between fifteen and seventy pounds, died on a scale sensitive to 1/16 of an ounce. No unexplained weight change appeared at death. MacDougall concluded that animals lack souls. He also acknowledged that the experiments were “vitiated by the use of two drugs” and that ideal conditions would require naturally weakened animals. He published this too.
MacDougall’s scale was sensitive to 2/10 of an ounce (about 5.6 grams), precise enough to detect his reported weight loss but also sensitive enough to register a breath, a draft, or a colleague’s stethoscope. Three of his six cases showed anomalies he could not cleanly explain.
What Clarke proposed
Augustus P. Clarke published a rebuttal in American Medicine in May 1907, one month after MacDougall’s paper appeared in the same journal.
Clarke pointed out that at the moment of death, the lungs stop cooling the blood and body temperature rises. That thermal shift triggers a surge of perspiration across the skin surface, which evaporates. Evaporated sweat is weight the scale loses, consistent with a ¾-ounce drop. No soul required.
Then Clarke turned to the dogs. They do not sweat through the skin; they regulate temperature through panting. At death, the same thermal surge would occur, but the sweat response would not follow. MacDougall found no weight loss in dogs not from an absence of souls but from an absence of sweat glands.
MacDougall’s counter-argument was that without circulation, blood cannot reach the skin surface in the first place, so no sweat surge could occur. The exchange continued in the journal through at least December 1907. Neither side produced the controlled replications that would have settled it.
Sanctorius of Padua had already established the mechanism. He spent thirty years in the early seventeenth century weighing himself continuously to track what he called “insensible perspiration,” the invisible, constant loss of mass through breath and skin, and published in 1614. By 1907, evaporative weight loss from the body was settled physiology. Clarke’s argument was that this is what MacDougall’s scale had measured.
From ¾ ounce to 21 grams
The New York Times story ran on March 11, page five, beneath the fold. By afternoon, other papers had it. Within days, the idea had traveled far enough to detach from its source.
MacDougall reported in imperial units, standard American practice in 1907. Three-quarters of an ounce converted to 21.262 grams. Secondary summaries dropped the decimal. Summaries of summaries dropped the fraction. The whole number acquired the specificity that numbers acquire when they stop looking like approximations.
The cultural carrier that fixed it permanently was Alejandro González Iñárritu’s 2003 film, starring Sean Penn, Naomi Watts, and Benicio del Toro. The film’s title is its only direct reference to MacDougall. The claim does not appear as dialogue or as a title card during the film itself. The three intersecting storylines carry the soul-weight idea as metaphor: the weight of a life against the lives it alters when it ends. The film received major awards nominations and ran for months in wide release. After 2003, the phrase “21 grams” required no explanation.
The “21 grams” figure was never in MacDougall’s paper. He reported three-quarters of an ounce. The gram conversion entered secondary accounts, and the 2003 Iñárritu film sealed it in popular consciousness, even though the film itself never states the claim in dialogue or as an on-screen caption.
What science has and hasn’t done since
No true replication of MacDougall’s experiment has been conducted. Mainstream science concluded within a few decades that measuring a hypothetical soul with a mass scale is physically incoherent, and closed the question.
In 2013, Jimo Borjigin’s team at the University of Michigan published a study in PNAS showing heightened gamma oscillations in rats during cardiac arrest, in exactly the brain regions responsible for visual experience and information integration. Interhemispheric communication spiked at death. A follow-up human study published in 2023 in Scientific Reports documented similar patterns in four dying patients. The sample is too small for strong conclusions, but both studies confirm that the dying brain does not simply switch off: organized activity continues whose relationship to experience remains unresolved.
Sam Parnia’s AWARE study (2008-2014) tracked 2,060 cardiac arrest events across multiple hospitals. Of 140 survivors who completed interviews, 9% had experiences qualifying as near-death experiences, and one case produced verified veridical perception during the arrest itself: a patient who accurately described specific events and equipment sounds during a period of documented isoelectric brain activity. No weight measurements were taken. Parnia’s work addresses consciousness and memory, not mass. But it sits in the same long argument MacDougall entered in 1907.
The older tradition behind this question is traced in When Breath Became Soul.
What the scale cannot settle
The Egyptian who placed a heart on Ma’at’s scales, the Christian artist who carved a weighing scene on a cathedral tympanum, the Quran’s author who promised judgment by the weight of a mustard seed: all were working on the same problem MacDougall brought to a hospital in Haverhill. Whether a life leaves something behind, whether what made a person who they were is a substance that persists or a pattern that dissolves, is the question no scale has settled.
MacDougall’s method was inadequate. His sample was too small, his calibration inconsistent across cases, and Clarke’s sweat objection remains technically unanswered. Two of the six cases were discarded for reasons he acknowledged. The one clean result cannot carry the weight he assigned it.
Every language that tried to name the soul named it as breath. Every culture that built a systematic picture of death built something that weighs or judges what a life amounts to. MacDougall tried to give that ancient project a number. He failed, and the project outlasted his number by more than a century. In 2023, four dying patients in a university hospital showed the same gamma oscillation surge that rats showed in 2013. No weight was measured.
