1. Introduction: A Word That Says Too Much and Too Little

“Shamanism” has long been a loose label. In everyday speech it blurs into “sorcery,” “magic,” and “the occult”; in scholarship it has served, at various times, as a catch-all for any dealings with spirits whatever. A word that can mean anything says nothing: to state of some practice that it “was shamanism” is almost always true and almost always empty. What is needed is not a better definition to legislate with, but an instrument to measure with — a small set of independent axes on which any given practice can be located, so that two practices bearing the same name can be told apart, and two practices bearing different names can be recognized as neighbours.

This paper assembles such an instrument from the classical literature, and is written to stand alone as a compact review. The sequence is as follows. Section 2 states Eliade’s strict definition — shamanism as the technique of ecstasy — and the distinction it draws between ecstatic soul-flight and possession. Section 3 defends that definition against its ethnographic critics by clarifying its logical status: an ideal type in Max Weber’s sense, whose value lies precisely in isolating the underlying variables. Section 4 takes up the most widely diffused typology in the popular literature, Joseph Campbell’s contrast of hunting and planting religions, and shows why it must be decomposed rather than adopted: it folds together variables belonging to different levels of description. Section 5 introduces the second axis, I. M. Lewis’s distinction between central and peripheral possession, with a stated safeguard: the description of position is taken, the functionalist explanation of motive is not. Section 6 populates the resulting plane from the best-documented ancient material, the Greek cases assembled by Dodds, and finds one cell vacant. Section 7 states, with Karl Jaspers, what the coordinates deliberately do not measure — the distinction between state and office. Section 8 presents a text that deserves to be better known: the sustained essay on the shaman in Henri Ey’s posthumous Naissance de la médecine, with its normative resolution of the “sick shaman” controversy and its insistence on healing as the centre of the shaman’s office. The application of the whole instrument to a single historical field — the two shaman-queens of early Japan, Himiko and Empress Jingū — is carried out in a companion paper[16].

2. Eliade: Shamanism as the Technique of Ecstasy

Before Eliade, “shamanism” was either dismissed as northern superstition or stretched to cover magical practice in general. Eliade cut the knot with a single formula[1]:

Une première définition de ce phénomène complexe, et peut-être la moins hasardeuse, sera : chamanisme = technique de l’extase.

[A first definition of this complex phenomenon, and perhaps the least hazardous, will be: shamanism = technique of ecstasy. (Payot ed., p. 22)]

Ecstasy here carries its etymological sense: ek-stasis, the soul’s going out of the body to travel to the upper or lower worlds. Every society has its medicine men and sorcerers; what marks the shaman off is this technique, together with its accompaniments — mastery of fire, magical flight, dealings with spirits without becoming their instrument. Eliade further distinguished two forms of commerce with the supernatural: the ecstatic form, in which the soul departs, and possession, in which the movement is reversed and a spirit descends into the body and speaks through its mouth. He located true shamanism in the ecstatic pole, where the human being remains the active master of spirits rather than their passive vessel, and he found its classical home among the Tungus, Yakut and other peoples of Siberia and Central Asia — the very word derives from Tungus šaman — while acknowledging its traces across North America, Oceania, and beyond. The shaman’s vocation, by heredity or by forced “divine election,” typically announced itself in a crisis of body and mind; what that crisis is, and whether it is pathology, is a question this paper defers to Section 8, where Henri Ey’s treatment of the “sick shaman” controversy is introduced.

3. The Ecstatic Type as an Ideal Type

Ethnography since Eliade has complicated his picture. Much actual Siberian practice is possession rather than soul-flight; the spirits are propitiated more often than commanded; fieldworkers warn that clean divisions of “shaman,” “priest” and “prophet” do not photograph reality[2][3][4]. If the pure ecstatic type is rare, is the definition refuted? No — provided its logical status is stated correctly. The ecstatic type is best read as an ideal type in Max Weber’s sense: a thought-construct that pushes the logic of a phenomenon to its most consistent extreme, whose value lies not in matching reality but in measuring how far, and in which direction, reality departs from it[5]. Read this way, Eliade’s purification of shamanism to a single point — the soul actively leaving the body under control — did exactly what a good instrument does: it isolated the underlying variables. There are two. The direction of movement: does the soul go out, or does the spirit come in? And the locus of agency: is the human being the master of the spirits, or their vessel? Possession thereby ceases to be a vague “trance with spirits” and acquires an exact definition — the form in which both variables are inverted. The rare and strange extreme of soul-flight serves as the light source that makes the familiar, easily overlooked structure of possession visible. On this reading the first axis of the coordinate system — the axis of technique, running from the ecstatic pole to the possession pole — is complete; and the religious-studies literature itself has come to defend precisely this ideal-typical construal of the term[6].

4. Decomposing Campbell: Levels of Description

The most widely diffused typology in the semi-popular literature is Joseph Campbell’s contrast, drawn in The Masks of God, between the religions of hunting peoples and those of planting peoples[7]. In hunting and pastoral societies the central figure is the shaman, whose authority rests on personal charismatic experience — trance, soul-flight, direct commerce with spirits — and whose paradigmatic partner is the Master of Animals, the supernatural owner of the game with whom the shaman negotiates on the community’s behalf. In planting societies the central figure is the priest, whose authority rests on institutional position and transmitted ritual knowledge: fixed sanctuaries, calendrical rites, the seasonal cycle of the Great Mother. The two types are not mutually exclusive; real societies mix them, and which one predominates is what characterizes a religious culture. As a first pass over almost any material, the scheme sorts usefully.

Taken as it stands, however, the dichotomy is wrecked by well-attested cases. The clearest in the ancient West is the Pythia at Delphi[8]: a woman through whose mouth the god speaks — possession-type in technique, personal and charismatic in mode — who nevertheless occupies an established, publicly sanctioned, inheritable institutional seat at the very centre of an agrarian civilization’s religious order. On Campbell’s scheme this combination should not exist: agrarian religion is supposed to be priestly, formalized, incompatible with the charismatic and personal mode of possession. The same “impossible” coordinate is occupied, at the other end of Eurasia, by Himiko, the third-century queen of the Wa confederation, whose kidō arose inside an agrarian society and yet was possession-type and central; her case, and its exact opposite in Empress Jingū, are analysed in the companion paper[16].

The diagnosis is that Campbell’s dichotomy folds together variables that belong to different levels of description. Four can be distinguished: the technique (ecstatic or possession — Eliade); the seat of authority (central or peripheral — Lewis, Section 5); the mode of subsistence (hunting-pastoral or agrarian — Campbell); and the lineage of the tradition (continental-northern or indigenous, or whatever the local genealogical question may be). The first two describe the religious office — how a society organizes the position of its religious functionary. The last two describe the society that hosts the office. Campbell’s types tacitly assume that the four always correlate: hunting = ecstatic, peripheral, northern; agrarian = liturgical, central, indigenous. The Pythia — and Himiko — prove the correlation is historical, not logical.

The decomposition also exposes an asymmetry. The hunting type survives it as a point with values on the decomposed axes (ecstatic technique, hunting-pastoral subsistence). The planting type does not: its substance is the priestly execution of prescribed liturgy, and the technique axis — which presupposes ecstasy of one form or the other — simply does not apply to it. One of Campbell’s types is a point on the plane; the other lies outside the axis. This is why his dichotomy cannot be erected as a third axis alongside technique and seat: it is not a variable but a bundle, and its familiar plausibility as a panorama rests on compressing four variables’ historical correlation into one name. What this paper calls the coordinate system is the space left after the levels are separated: at the level of society, subsistence and lineage, recorded as annotations; at the level of the office, technique and seat of authority, erected as axes; and at the level of the individual, the state — the quality of experience — which the system deliberately does not measure (Section 7).

5. Lewis: Central and Peripheral Possession

The second axis comes from I. M. Lewis’s Ecstatic Religion[9]. Central possession is possession by established, morally authoritative spirits — ancestral or divine guardians of the community’s order — incorporated into the mainstream and legitimating the existing political-religious structure. Peripheral possession is possession by marginal, often amoral or foreign spirits, borne by socially peripheral persons, and functioning as an oblique, case-by-case channel to voice or resources otherwise unobtainable. The decisive advantage of the distinction is that it is not tied to any mode of subsistence: Lewis’s material runs across pastoral, agrarian and mixed economies, and both forms can coexist in one society.

Lewis’s own ethnography gives the distinction flesh. The textbook case of peripheral possession is the saar possession of married women in patriarchal Somali society: a woman without public voice falls ill, possessed by an amoral foreign spirit; the spirit, through her mouth, demands perfume, clothes and delicacies of her husband; the cure is admission to a women’s possession cult, and the affliction recurs — the demand that could never be made directly is made obliquely, the strategy of the powerless (the zar and bori cults are cognate). The paradigm of central possession is the mhondoro mediumship of Shona society: the royal ancestor spirit descends not on the chief but on a dedicated medium, who adjudicates rainfall, land morality and succession from an established institutional seat — and, crucially, the authority belongs to the spirit and the seat, not to the medium’s person, so that when a medium dies the same spirit takes a new one and the office continues. An inheritable office irreducible to any individual’s state: a structure to which Section 7 returns. Haitian vodou, finally, shows that centrality is relative to the reference community: peripheral from the standpoint of the state, the descent of the lwa is the central moral mode within the peasant congregation.

One safeguard completes the axis. From Lewis this paper takes the description of position only, not the functionalist explanation of motive. The caution is empirical as well as methodological: Carmen Blacker, examining possession affliction in Japan — where the afflicted are likewise predominantly young married women — declined the analogy with saar: the quiet exorcisms of temple and healer offer the patient no stage and no goods, and the spirits’ demands profit her nothing[10]. The peripheral position of such bearers stands; the strategic motive does not travel. Wherever the axis is applied, the claim “peripheral” should be read structurally — an authority not settled in an established seat — and nothing more.

The coordinate system is now complete: a plane spanned by the axis of technique (ecstatic–possession) and the axis of the seat of authority (central–peripheral), both properties of the office, with the society-level variables (subsistence, lineage) recorded as annotations of a located point rather than as axes.

6. The Plane Populated: The Greek Witness, and a Vacant Cell

An instrument is tested by using it. The best-documented ancient field is the Greek one, on the ground assembled by Dodds in The Greeks and the Irrational and its appendices[8], and there three of the four cells are filled. The Pythia at Delphi — possession-trance in an established, publicly sanctioned seat engaged with the order of the city-states — is the pure case of central possession. The wandering “god-men” of the northern legends — Abaris, Aristeas, Hermotimos, Epimenides — whose souls travel while their bodies lie as dead, fill the ecstatic-peripheral cell. And the maenads of the biennial mountain rites, female, periodic, outside the institutional centre, are the ancient world’s textbook case of a peripheral possession cult, their spontaneous seizure tamed by ritual organization. That Meuli’s derivation of the god-men from Scythian shamanism has since fallen to Bremmer’s critique[11] does not touch the location: typology is not genealogy, and when where they came from falls, what they were stands — on the broader anthropological basis of the free-soul, which binds the ecstatic state to no particular lineage.

The fourth cell — ecstatic and central — appears vacant in Greece, and equally vacant in the Japanese material examined in the companion paper[16]. One hypothesis may be recorded: ecstasy is in principle a private event, reaching the community only by the ecstatic’s own report, whereas possession is performed in public, heard as an altered voice, and communally verifiable after the event; a central seat demands public verifiability, and the unverifiable resists central institutionalization. The test of that hypothesis belongs to a third civilization — and the coordinate system is offered precisely so that such a test can be run.

7. State and Office

What do the coordinates deliberately not measure? The answer is owed to Karl Jaspers[12]: the state — the quality of experience that occurs in an individual — and the office — the institutional seat that a society designs — are different dimensions, however similar their surface expression. Both axes of the plane are properties of the office; the state varies independently of them, and the Japanese ethnographic record supplies the two extreme cases that prove the independence. The blind mediums of northern Honshū, the itako, retain the office without the state: researchers from Hori onward decline to count them true shamans — no initiatory crisis, no election by dream or sudden possession, and what passes for trance among them is, on careful observation, imitation — yet the office persists as a full institution, with apprenticeship, initiation, implements and fees[10]. The founders of the modern Japanese new religions — Nakayama Miki, Deguchi Nao, Kitamura Sayō — present the state without the office: the full shamanic biography of sickly eccentricity, shattering election and transformed personality, but no prepared seat; the seat has to be built afterwards[10]. State and office are thus two independently varying quantities. It is this independence that entitles a typology of the office to proceed without any access to the practitioner’s inner experience — a point of no small consequence for historical cases, where no such access will ever be had.

8. Henri Ey’s Shaman Essay: The “Sick Shaman” Controversy and the Healing Office

The last instrument comes from psychiatry. Henri Ey (1900–1977), the theorist of organo-dynamic psychiatry, worked until shortly before his death on a history of the origins of medicine; edited by Henri Maurel, it appeared posthumously as Naissance de la médecine (1981), and it contains a sustained twenty-page treatment of the shaman (CREHEY edition 2017, pp. 56–76) that has never been translated and is little read even in French[13]. A full presentation and critical discussion of the text is given in a companion guide[17]; here what bears on the review is taken up.

Recall the question deferred from Section 2. The shaman’s vocation typically announces itself in a crisis of body and mind, and early ethnology read that crisis as pathology — “arctic hysteria,” a nervous disease of the far north. Around this reading a controversy ran for over a century, and Ey, as a participant, mapped it: on the pathological side Krivoshapkin (1861), Wilkens (1887), Bogoras (1910), Czaplicka (1914), Ohlmarks (1939), Kroeber (1952), Linton (1956), La Barre (1967) and Devereux (1970); against them Opler (1936), Ackerknecht (1941), Bouteiller (1950), Eliade (1951) and Hoffmann (1968)[13].

What deserves attention is how Ey adjudicates. He first registers the empirical rejoinder — observers agree that most shamans behave normally outside the trance — and then sets beside it a stronger argument of a different kind: by definition the shaman is not sick; quite the contrary, for he is a member of a cultural group governed by a normative institution, and he conforms to its normativity. The argument is neither statistical nor clinical but normative, and it has a sharp reverse face: the magician, the healer, the shaman could be counted “sick” only by assimilating them to the very patients they heal — and in that world no such assimilation is possible, because all members of the group stand equally under one status, which is not disease but Mal, harm and ill. A line between the sick and the well becomes drawable only once the concept of natural disease exists. In the magico-animistic world, Ey’s central thesis runs, there is no category of “disease” at all; shamanism stands before medicine and against it at once — ante-médecine and anti-médecine[13].

Ey’s text also touches the first axis of this review, from an unexpected side. German-language psychopathology and ethnology had forged their own pair of concepts at the Homburg symposium of May 1968 — Ergriffenheit, the state of being seized as by grace, and Besessenheit, possession — recorded in the volume edited by Jürg Zutt[14]; and Ey, describing the shaman’s rapture, cites precisely this pair in a note[13]. The pair belongs to the level of the state — it distinguishes qualities of experience, not organizations of the office — and so it cannot be identified with the technique axis. But the cut falls in the same place; that two independent traditions, one phenomenological and one typological, drew their boundary at the same point is corroboration that the axis is no school’s private speculation. And there is a methodological lesson in the fact that Ey left the vocabulary in a footnote: in his own text the shaman’s trance is described now as migration, levitation and ascent, now as “possession and dispossession,” the two modes fusing in a single drama. To possess a distinction and to erect it as an axis are two different things.

Two further passages deserve notice. First, Ey noticed what Durkheim did not say: Les Formes élémentaires never, so far as he could see, uses the word “shaman,” and contains no chapter or paragraph on a person, office or rite of healing — and he explained the silence: the Durkheimian clan is conceived as a purely animistic organization, a circulation of magical force, in which body, life, death and disease can appear only as incidents of the struggle with evil; within such a construction there is no room to erect the healing office as a level of its own[13]. Whatever the ethnographic merits of Ey’s rereading of the Arunta alatunja, the form of the diagnosis — a theory’s blind spot explained by its folding of levels — is exactly the form applied to Campbell in Section 4. Second: Ey placed healing at the centre of the shaman’s office. His shaman holds “the powers of divination and healing,” and “his principal function is to heal”; rain-making, the warding of calamity, the settlement of thefts and quarrels are arranged around that centre[13]. This determination from the side of content does work that the axes of mode and position cannot: it cuts specialized figures out of the broader concept. An oracular office without healing — a “shaman-queen” who divines but does not cure, of which early Japan offers a dual pair — stands out, against Ey’s criterion, as a specialization rather than a full instance of the office; the case is argued in the companion paper[16].

A last convergence reaches the present. The diagnostic guidelines of ICD-11 build cultural exception into their very structure: trance and possession states “should only be considered to be features of a mental disorder when they are involuntary and unwanted, not accepted as a part of a collective cultural or religious practice,” and productive of significant distress or impairment[15]. The logical structure is Ey’s: the criterion is located not in the description of the phenomenon itself but in the position the phenomenon occupies within the normative system of its group. No influence need be supposed — Ey’s text was hardly read even in France — and the convergence is the more instructive for that: it attests that the clause is not a concession to relativism but the consequence of an argument from normativity.

9. Conclusion: What a Coordinate System Is For

The instrument assembled here is deliberately small: two axes at the level of the office — technique (ecstatic–possession, Eliade read as ideal type) and seat of authority (central–peripheral, Lewis read as description of position) — with society-level variables recorded as annotations, the individual’s state excluded on Jaspersian grounds, and one determination of content, Ey’s healing office, held in reserve for cutting specializations out of the broad concept. Its uses are three. It tells apart practices that one name has fused: “shamanism” at Delphi and “shamanism” on the steppe are different cells of one plane. It recognizes neighbours that different names have parted: the mhondoro medium, the Pythia, and the institutionalized oracles of other civilizations occupy one cell under three vocabularies. And it converts vague incongruities into questions decidable axis by axis — which is what any coordinate system is for.

The review has also marked its own limits. The plane does not measure the quality of anyone’s experience; it does not decide genealogies, and survives their fall; and one of its cells is vacant in the two civilizations examined, a vacancy for which a hypothesis — public verifiability as the price of a central seat — awaits its test elsewhere. The full trial of the instrument on a single historical field, where two figures bearing one label land at opposite poles and the divergence proves to be design rather than noise, is the burden of the companion paper[16].

Notes and References

Citations are to original-language editions; English translations are noted where they exist.