Note 050

Where the Human Pillars Are Said to Be

By KWAIDANOTE · Published September 28, 2026

A Skeleton and a Newspaper

In 1925 human remains were found during work below one of the turrets of the imperial palace in Tokyo, and the newspapers reported what everyone expected them to report: a human pillar, a person buried alive in the foundations to hold the building up.

The folklorist Minakata Kumagusu wrote an essay in reply. Our note on the practice itself sets out what the rite was supposed to be; this one is about his essay, because of the form he chose for it. He did not argue. He made a list — of every place in Japan said to have a body under it, and, in each case, of exactly who said so.

That second half is the part worth having.

What Was on the List, and Who Was Behind It

The bridge at Nagara in Settsu, where the story is oldest and most often repeated: he follows an illustrated encyclopaedia for the outline, and takes the particulars — the year 812, the name of the daughter — from a topography of the province.

A village in Rikuchū, where a man was buried with his horse: from a paper in an anthropology journal, credited to a named scholar.

Ōzu castle in Iyo, where a woman called Okame was chosen by lot: from a man of the town telling him directly, and from a letter sent to him by the compiler of the county gazetteer.

Matsue castle, where the girl taken was one who passed by every night chanting beautifully: from another named informant, and from a published collection of Japanese legends.

Inawashiro castle and Himeji: from two Edo-period miscellanies, one of them the enormous notebook of a daimyo.

Wakayama castle, where the woman was called Otora: from something an old man told him when he was a child.

Set out like that, the tradition stops being a tradition and becomes a bibliography — gazetteers, journal articles, private letters, personal memory, arranged in descending order of checkability, with the weakest at the end and marked as weakest.

How the Victim Is Chosen

One thing emerges from the list that no single story shows, and it only becomes visible when the cases are set side by side: the selection methods do not match.

At Ōzu the woman is picked by lot. At Matsue castle the girl taken is chosen because she went past every night chanting beautifully — that is, because she had been noticed. At Nagara, and again in the separate legend of the great bridge at Matsue, which our note on Lafcadio Hearn takes up, the choice is fixed by a rule about clothing, so that it falls on whoever is wearing the wrong garment when the rule is applied. And in the stage version of Nagara, written three centuries later, a man volunteers.

Lot, being noticed, a rule about clothes. Not one of them is a procedure for finding the right person; each is a procedure for making the choice look like nobody’s. The volunteer belongs to the play, and that is the point about the play: it is the one version in which somebody decides.

Lot, talent, volunteering, accident. These are not variants of one rite. They are four different procedures for deciding who may be killed, and each of them puts the responsibility somewhere else — on chance, on the victim’s own visibility, on the victim’s own offer, on a rule nobody made for the occasion. That is the sort of pattern a list produces and a single story never can.

The Oldest Case Already Contradicts Itself

The example he puts first is the artificial island built in the twelfth century in what is now Kobe.

One war chronicle says a boy on a white saddle was put into the works. Another says that stones inscribed with sutras were used in place of a person. Minakata sets the two side by side and points out that they disagree.

That is a useful thing to have at the top of a list. The earliest and grandest instance is a disagreement between two texts about whether it happened at all, and everything after it in the list is younger.

The Bridge and the Play

Nagara has a further complication, and it is the kind this site keeps running into.

In the old telling the man chosen is the father, and the sting is in what happens next: his daughter, married into another province, keeps silent about it and is betrayed by her own husband for the silence.

In 1913 Okamoto Kidō wrote a play about the same bridge, revised it in 1928, and had it staged at the Kabuki-za in April of that year. In the play the person trying to prevent the sacrifice is the son, hurrying home from the capital to stop his father being used.

The rearrangement is small and it moves the whole weight of the story. The old version is about a woman who has to live afterwards. The play is about a young man who might arrive in time.

What the Castles Say About Themselves

Matsue castle publishes the story on its own website. It appears under a heading for strange tales about the construction, beside other ghost stories, told as something people say rather than something the castle asserts. That is a fair way to handle it and it is worth noting that the castle does handle it that way.

For Maruoka castle, where the legend of a one-eyed woman named Oshizu is widely repeated, I was not able to obtain the city’s own primary statement, and so this note does not report one.

Bones That Were Actually Found

Two cases in Japan are cited where human remains were found and the finding was reported rather than merely told.

Behind a temple in Niigata prefecture, in 1937, a man’s skeleton was found in a large jar, in a seated posture, in fill behind the building; the case is described as confirmed. In Hokkaido, in 1970, repair work on a railway tunnel opened in 1914 exposed a standing skeleton in the wall, and more remains near the entrance.

Neither of these belongs to the castle-and-bridge tradition. The second is generally understood as the disposal of a labourer who died under the brutal conditions of the work gangs, which is a different kind of horror and a documented one.

And the famous cases — Nagara, Matsue, Himeji, Inawashiro, Ōzu — have not been confirmed by excavation. None of them.

What the Category Is Actually For

One reading of all this is that some of the legends must be true and we simply have not dug in the right places.

The essay suggests something else, and the occasion of its writing is the proof. A skeleton came out of the ground under a palace turret, and within days it had a story attached to it that was older than the building, more interesting than the likely truth, and available in stock. The list Minakata assembled is not a survey of what is under Japanese buildings. It is a survey of what people say when something is found under one.

Sources

Minakata Kumagusu 南方熊楠, “On human pillars” 人柱の話, first published in a psychology journal in September 1925 and collected in his essays in 1926; the list of sites with his attributions, including a provincial topography for the Nagara particulars, an anthropology journal for Rikuchū, private letters and oral testimony for Ōzu, a published collection of legends for Matsue, and Edo-period miscellanies for Himeji and Inawashiro.
Okamoto Kidō 岡本綺堂, Nagara no hitobashira 長柄の人柱, written 1913, revised 1928, staged at the Kabuki-za in April 1928.
The city of Matsue’s own account of the castle, for the construction tales as it publishes them.
On remains found and reported: the 1937 discovery behind a temple in Niigata prefecture, and the 1970 discovery during repairs to a Hokkaido railway tunnel.