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 known record: the Chōkyū era (1040–1044), in the tale collection Dainihonkoku Hokke Genki, compiled by the monk Chingen.
• 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, a woodblock print and local oral tellings in other parts of Japan.

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 some tellings, 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 known 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. The second collection is usually dated to around 1120.

A Japanese encyclopedia entry gives a tighter date for the first collection than the round figure of “about 1040” often quoted. It identifies the work as compiled by the monk Chingen in the Chōkyū era, which ran from 1040 to 1044, and names the Muro district tale as the earliest record of the legend. That turns an approximate year into a four-year window, and it attaches the oldest text to a named Buddhist compiler rather than to an anonymous tradition.

Those dates still come from reference works rather than from inspection of the manuscripts themselves, and should be read as the standard scholarly dating, not as a date read off a colophon.

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.

The picture scrolls are not a single object. The Wakayama Prefectural Museum holds six scrolls of the Dōjōji Engi in five distinct versions, and the museum describes how their pictures and texts differ. One version leaves out the scene of the Lotus Sutra memorial service. Another already draws a serpent into the scene where Anchin and Kiyohime first meet, so the ending is foreshadowed from the start. A third, the C version, is written in colloquial phrasing and preserves how the scroll was used in etoki — picture-telling, in which a narrator explains the images to an audience — during the Edo period.

The legend also has a documented local geography beyond the temple. Wakayama Prefecture’s official history site places the house where the monk asked for a night’s lodging in what is now the Nakahechi area of Tanabe City, and names the householder as Masago no Shōji Kiyoshige, head of the Masago family in its third generation and, in this telling, Kiyohime’s father. The same site lists places in the area that local tradition identifies as her birthplace and her grave, and suggests a route for visiting them.

Finally, the story did not stay in Kii Province. The folklore database maintained by the International Research Center for Japanese Studies includes local tellings of Anchin and Kiyohime, recorded in sources dated between 1918 and 1989. Alongside many records from Wakayama, it includes versions gathered in Tochigi, Fukushima and Miyagi, far from Dōjōji.

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 other reference accounts 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.

What the sources add

Read together, the dictionary, museum, prefectural and research-institute sources sharpen the picture in three ways.

The first is dating. The earliest record moves from “around 1040” to a compilation in the Chōkyū era, 1040–1044, by a named monk. That is a small correction, but it anchors the legend to a specific Buddhist compiler working in the mid-eleventh century, which fits the story’s function as a teaching tale rather than a report of an event.

The second is variety. Accounts of the legend usually speak of “the” Dōjōji Engi scroll, while one regional museum alone holds five versions with different scenes and different wording. One of them was written for oral performance, which shows that the legend was carried by narrators working from images as well as by printed or dramatic texts.

The third is place, and here the sources diverge from the oldest text. The Heian tale gives an anonymous widow in Muro district; the prefectural history site gives a named father, a household in the Nakahechi area and a birthplace and grave that visitors can walk to. These sites belong to local tradition built up around the legend, not to the earliest written record, and the source presents them as traditions rather than as proof that the figures existed.

Meanwhile the folklore records show that the story also travelled as oral tradition, retold in regions far from the temple and still being recorded by folklorists in the twentieth century.

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.