Kiyohime: The Serpent Woman of Dōjōji Temple

On a kabuki stage, a dancer in the guise of a shirabyōshi performer named Hanako drifts closer and closer to a newly cast temple bell, and keeps dancing for well over an hour. The piece is Kyōganoko Musume Dōjōji, first staged in Edo in 1753, and it assumes the audience already knows why that bell is the centre of everything: at this temple, the story goes, an earlier bell was destroyed by a great serpent. The serpent, in the tale, had once been a woman.

That woman is Kiyohime, and the story behind the dance is older than the dance by roughly seven centuries.

Quick facts

• Verdict: traditional folklore. Kiyohime comes from a Buddhist setsuwa (説話) — a didactic tale of karmic retribution — recorded in Heian-period collections, not from a documented historical incident.
• Name: Kiyohime; in Japanese the story is usually called the Anchin–Kiyohime legend, after its two central figures.
• Type: a religious tale that became a stock subject of classical Japanese performing arts and visual art.
• Earliest confirmed record: around 1040, in the tale collection Dainihonkoku Hokke Genki.
• Where it spread: attached to Dōjōji temple in Kii Province, in what is now Wakayama Prefecture, and carried far beyond it by noh, kabuki, jōruri puppet theatre, picture scrolls and a woodblock print.

What the story says

A monk travelling on pilgrimage to Kumano stops for the night at a lodging, where a young widow falls in love with him on sight. He puts her off with a promise to call on her again on his way back, and then leaves. When she realises she has been deceived, she sets off after him. At the Hidaka River her pursuit turns monstrous: she takes the form of a serpent and crosses the water. The monk flees to Dōjōji, where he is hidden beneath the temple’s great bell, but the transformation has gone too far to be undone by sanctuary. The serpent coils around the bell, and he dies inside it — in the English-language telling, she strikes the bell with her tail and then breathes flames fierce enough to melt the metal.

Two things are worth noticing about this summary. First, it is a religious story with a moral engine: desire, deception and consequence, told in collections compiled to illustrate Buddhist teaching. Second, in the earliest surviving versions neither figure has a name at all. The monk is simply a monk and the woman simply a young widow; the names Anchin and Kiyohime were attached to them much later, by writers working centuries downstream of the original text.

Earliest confirmed record

The oldest known version is in the Dainihonkoku Hokke Genki, a Heian-period collection of Buddhist tales, in a chapter about a wicked woman of Muro district in Kii Province. A nearly identical text was transmitted into the Konjaku Monogatarishū, in the third tale of its fourteenth volume; the older version is written in Chinese-style prose, the later one in Japanese reading order. English-language reference sources date the first collection to around 1040 and the second to around 1120.

Those dates come from secondary reference sources rather than from inspection of the manuscripts themselves, and should be read as approximations.

What is documented

What can be documented here is the textual and theatrical record, not the events.

The transmission is clear and traceable: the tale exists in the two Heian collections, and it is tied to the founding tradition of Dōjōji in Kii Province, modern Wakayama Prefecture. The naming history is equally traceable. The monk’s name, Anchin, first appears in the Genkō Shakusho of around 1322 — some three centuries after the earliest text. The name Kiyohime is later still, first appearing in the jōruri play Dōjōji Genzai Uroko, premiered in 1742. The woman whose name now headlines the legend was anonymous for most of the legend’s history.

The performance lineage is also documented. The noh play Dōjōji is a sequel to the tale: a shirabyōshi dancer — in truth a vengeful spirit — arrives at the bell-dedication ceremony at Dōjōji in Kishū (Kii Province), dances, and leaps into the bell; when the bell is raised she appears transformed into a serpent, breathing fire, until the monks’ prayers overcome her and she plunges into the river and vanishes. The play is famous for its ranbyōshi, a drawn-out war of nerves between the dancer and the hand-drum player that finally breaks into a rapid dance, and for a costume change executed on stage.

The kabuki dance follows the same premise. Kyōganoko Musume Dōjōji premiered in the third month of 1753 at the Nakamura-za in Edo, with lyrics by Fujimoto Tobun, music by Kineya Yasaburō I and choreography by Nakamura Denjirō I. A new bell is being dedicated at Dōjōji where a great serpent once destroyed the old one, and the shirabyōshi Hanako appears and dances her way toward it. In practice the drama matters less than the dancing: the piece runs over an hour and is regarded as one of the summits of the kabuki dance repertoire.

What is contested

Dating the visual and theatrical versions is where the sources stop agreeing.

The Dōjōji Engi picture scrolls — two scrolls, designated an Important Cultural Property — are assigned by temple tradition to 1403 and to the brush of Emperor Go-Komatsu. Modern examination points instead to the late fifteenth or early sixteenth century, and English-language reference sources simply place them in the Muromachi period, around the fifteenth century. There is no basis for treating any one of these as settled.

Authorship of the noh Dōjōji is similarly unresolved. It has been credited to Kan’ami and Zeami and also to Kanze Kojirō Nobumitsu, who died in 1516; one account holds that the play is a condensed, restructured version of the earlier Kanemaki, rebuilt around the ranbyōshi sequence. Attribution in early noh is frequently like this, and the safest statement is that the play exists and its author is disputed.

The largest contested point, though, is the one readers outside Japan most often get wrong. Nothing in the record establishes that any of this happened. There is no documented killing at Dōjōji behind the legend — only a Buddhist tale, copied between collections, given names by later writers, and then handed to some of Japan’s greatest dramatists.

Adaptations

The kabuki dance that opened this article is one node in an unusually dense adaptation network. Alongside Kyōganoko Musume Dōjōji (1753) stand the jōruri plays Dōjōji Genzai Uroko (1742) and Hidakagawa Iriaizakura (1759), the noh Dōjōji, the Dōjōji Engi picture scrolls, and a woodblock print by Tsukioka Yoshitoshi published in 1890. Each medium kept a different part of the story: the noh kept the bell and the kabuki kept the dance.

How it is discussed in Japan today

Ocsoku, a Japanese occult blog, has covered the subject in one post, published in 2018.