Shamans of the Shivaliks

“Transported with our Munihood we have pressed on into the winds: you therefore, mortal men, behold our natural bodies and no more.”

— Ṛgveda X.136, the hymn of the long-haired muni, trans. Griffith (1896)

“Oral traditions are documents of the present, because they are told in the present.”

— Jan Vansina, Oral Tradition as History (1985)

The Himalayan foothills are remembered through texts and temples, and the archive has a bias: it was written by the literate, for the purposes of the literate. Beneath that recorded layer runs an older stratum — village oracles, possession mediums, deity-speakers, and healing specialists — which survives in fragments while its pre-colonial form goes unrecorded. The spine that preceded this paper set out a section plan and a list of sources to verify. This paper does the work the spine promised: it verifies the sources, reconstructs what the evidence will bear, grades each claim by confidence, and marks the line beyond which the Institute will not go.

The method is the one the Institute applies wherever the written record belongs to the administrator and the practice belongs to the people: read the record against its own purpose; treat oral tradition as an archive with known failure modes; and accept a reconstruction only where three strata of evidence agree — a textual trace that predates colonial ethnography, a vocabulary the record could not have introduced, and an institutional fossil that survived independently of memory. Call this the Stratified Evidence Test. Anything that passes it is claimed. Anything that fails it is named as a hypothesis or dropped.

Three refusals govern the paper. It does not argue that the foothill stratum is “timeless”: survival is not proof of age, and the word shaman is used in the analytic sense given it by cross-cultural anthropology — a specialist who enters an altered state to deal with unseen agents on behalf of others — not as a claim of Siberian, Vedic, or hunter-gatherer descent. It does not depend on the entheogen question, which the Institute argues elsewhere; nothing here stands or falls with it. And it does not film anything: the documentary treatment in Section VI sets rules for a film that has not yet been shot, so that the film cannot later be accused of having staged what this paper could not prove.

Disclosure. The documentary treatment originates with a member of the Editorial Board whose family comes from the Hoshiarpur corridor discussed below. That connection is an access condition and a declared bias. It is not evidence, and nothing in this paper rests on private family memory.

---

I. The Foothills as a Borderland

Geography is the first method. The Shivaliks — the outer Himalaya, a young and low range of sandstone and conglomerate — separate the Punjab plain from the middle hills of Kangra, Chamba, Bilaspur, Mandi, and Kullu, and beyond them from the high country of Bashahr, Lahaul, and Kinnaur. Along their southern foot runs the kandi, the dry gravel belt cut by seasonal torrents (chos) whose floods so alarmed the colonial state that it legislated against them by name in the Punjab Land Preservation (Chos) Act of 1900. Hoshiarpur district, which stretches from the Doaba plain up into the Shivaliks, is the hinge of this geography: a plains district administered from Jalandhar and Lahore that looks, on its northern edge, into the hills.

The political history of the belt is a history of never being fully owned. The Katoch kingdom of Kangra and the raj of Chamba are among the oldest continuously attested polities in the subcontinent; Chamba’s copper-plate grants, edited by Vogel in 1911, run from the tenth century and record land given to temples and Brahmins by hill kings who answered to no plains emperor. The Mughals held Kangra fort within the Lahore province and knew the flame-shrine at Jwālāmukhī well enough for Abu’l-Fazl to note it and for Jahāngīr to visit it on his 1622 progress — but Mughal authority in the interior valleys was tributary, not administrative. Ranjit Singh took Kangra fort in 1809 and made the hill rajas his tributaries; the East India Company took the Jalandhar Doab and the hill country between the Beas and the Sutlej by the Treaty of Lahore in 1846, annexed Kangra and Kullu, and left the remaining hill states as princely states. From the plains’ point of view the foothills were taxed, garrisoned at chokepoints, and otherwise left alone. From the hills’ point of view the plains were a source of grain, cloth, and occasional armies.

Borderlands conserve what centres discard, and the mechanism is not romantic. It is administrative economics. A practice survives where the cost of suppressing it exceeds the revenue of the territory in which it is practised; James C. Scott’s account of the Southeast Asian hill massif as state-evading space is the general model, and it applies, with local corrections, to the Himalayan foothills. The corrections matter. Foothill society was not stateless. It had its own states — and, crucially, its own theistic polity. In Bashahr and the Shimla Hill States, in Kullu, Mandi, and the upper Beas, the village deity (devtā) was a juridical person: a landholder, a judge, a sovereign who travelled his domain in a palanquin, and who spoke through a human oracle (gur) in public consultations that adjudicated disputes, allocated water and pasture, and diagnosed misfortune. Peter Sutherland’s ethnohistory of Bashahr named this devtā kā rāj, government by deity, and showed that it prevailed from pre-colonial times and survived British paramountcy because the British, needing revenue and order, recorded the deity’s land and the deity’s fairs and left the deity’s speech alone.

This is the borderland’s real conservation mechanism, and it is the hinge of the entire paper. The oracle institution survived because it was anchored in tenure, and tenure is the one thing every regime records. A muāfī (revenue-free) holding in a devtā’s name, entered in a settlement record by a British deputy commissioner in 1883, is an institutional fossil of a consultation practice that the same deputy commissioner filed under “superstition.” The gazetteer preserved the skeleton while discarding the body.

One further correction to the refuge model is decisive for chronology. What the empires did not reach, reform did. No Mughal faujdār and no Sikh kārdār ever attempted to change what a hill peasant believed. The Singh Sabhā movement (from 1873) and the Ārya Samāj (from 1875) did — through print, schools, preaching tours, marriage law, and finally the Sikh Gurdwaras Act of 1925, which reorganised the custody of shrines across Punjab and removed the Udāsī and Nirmalā custodians whose lineages had carried the older pluralist practice. Harjot Oberoi’s The Construction of Religious Boundaries (1994) remains the standard account of this operation. The foothill belt, closer to the plains and served by the same print and rail, was more exposed to reform than the interior valleys; the Doaba villages on the Shivalik foot were most exposed of all. The gradient of survival therefore runs uphill: densest in Bashahr and Kullu, thinner in Kangra and Hamirpur, thinnest in Hoshiarpur. That gradient is itself evidence, and Section V returns to it.

---

II. The Surviving Practices

What follows is a survey, not an assertion. Each practice is listed with its zone, its human specialist, the earliest written record this paper has verified, and its evidentiary grade. The discipline is that survival is not proof of age: a living practice is a candidate, and its antiquity must be argued from the record, not assumed from its vitality.

Practice	Zone	Specialist	Earliest verified record	Grade
Village-deity oracle (gur) and devtā kā rāj: public consultation of the possessed oracle; deity as landholder and judge	Kullu, Mandi, Bashahr, Shimla hills	gur (oracle), kārdār (manager), pujārī	Hill-state land grants; Kangra and Simla Hill States gazetteers (1883–1910); Vidal 1988; Sutherland 1998; Berti 2001	Institution: HIGH (via tenure). Repertoire: MEDIUM
Assembly of devtās at Kullu Dussehra	Kullu	as above	Gazetteer-recorded tradition of Raja Jagat Singh’s installation of Raghunāth (mid-17th c.)	MEDIUM
Jāgar: night-long drummed séance in which a medium is possessed and speaks	Garhwal, Kumaon (eastern analogue)	bākī / ḍangariyā (medium), jagariyā (drummer-singer)	Berreman 1963; Alter 2008; Sax 2009	Institution: HIGH; foothill relevance: analogue only
Nāth-Siddh cult of Bābā Bālak Nāth at Deotsidh; goat offering (nurtured, not sacrificed)	Hamirpur–Bilaspur border; pilgrims from Una, Hoshiarpur, the Doaba	shrine custodians in the Nāth orbit; chelās	Briggs 1938 on the region’s Nāth cults; 20th-c. shrine record	Nāth stratum: HIGH. Present shrine economy: RECENT
Nuālā: nocturnal rite to Śiva with trance and animal offering	Gaddi communities, Chamba–Kangra	Gaddi ritual specialists	20th-c. Gaddi ethnography	LOW–MEDIUM
Guggā Pīr cult; Guggā Naumī in Bhādoṅ; snakebite protection; devotees’ chain-flagellation; Chhapar Mela	Doaba, Malwa, Haryana, hill fringe	bhagat carrying the chhaṛī; snake specialists	Temple’s recorded recitation of the Guggā epic (1884–1900); Ibbetson 1883; Rose 1911–19	Cult: HIGH by the 1880s; pre-1849 depth: MEDIUM
Sakhī Sarwar (Sultānī) cult; pilgrimage to Nigāhā	Central Punjab, Doaba	bharāī (hereditary minstrel-priests)	18th-c. rahit prohibitions (as read by Oberoi 1994); 1881 and 1891 census categories	Cult: HIGH; pre-colonial depth: ARGUED
Jaṭherā: clan-ancestor shrine at the village boundary, often at a jaṇḍī tree	Punjab villages, plains and kandi	family and clan	Rose 1911–19; Bhatti and Michon 2004	MEDIUM
Village exorcist-healer (siyāṇā); sweeping-and-blowing cure (jhāṛ-phūṅk); amulets (tāvīz)	Plains and foothill villages	siyāṇā, bhopā, pīr	19th-c. reform and census sources as read by Oberoi 1994; Kakar 1982	Practice: HIGH by the 1880s; pre-colonial depth: MEDIUM

Three observations follow from the table.

First, the practices are not one thing. The gur of Kullu is a public, juridical, community-owned institution; the siyāṇā of a Doaba village is a private specialist paid per case. The oracle speaks as the god; the healer speaks to the spirit. Both enter altered states, both deal with unseen agents, and both are “shamans” in the analytic sense; but their social positions differ, and their fates under reform differed. The public institution, protected by tenure, survived. The private specialist, protected by nothing but reputation, was the first casualty of the reformers’ pamphlets.

Second, the healing emphasis visible today may itself be recent. Sharabi and Shalev’s study of the deity Mahāsū documents a shift, within living memory, from ruler to healer: as state courts absorbed the deity’s judicial functions, consultations narrowed toward illness and affliction. What a documentary camera sees in a foothill shrine in 2026 — a queue of the sick — may be the modern face of an institution whose older function was constitutional. The healing is old. The centrality of healing may not be.

Third, the Nāth stratum ties the list together and is the one component that can be dated with confidence from a text the region itself preserves. Bābā Bālak Nāth at Deotsidh is a Siddh in the Nāth orbit; Gorakhnāth’s Kānphaṭā yogis held the great seat at Tilla Jogian on the Pothohar plateau and lesser seats throughout the Punjab hills; and the Nāth ascetic vocabulary — the dhūnī (sacred fire), the bhabhūt (ash), the mātrā (the verse-code of ascetic discipline) — runs through the Udāsī and Sanātan lineages that later custodianship disputes would try to pull apart. The dating text is the Siddh Goṣṭ, Guru Nānak’s dialogue with the Siddhas, preserved in the Guru Granth Sāhib at Ang 938–946. It is a sixteenth-century witness, fixed in the Ādi Granth of 1604, that Nāth yogis were the interlocutors a reformer had to argue with. Whatever else is uncertain, the Nāths were here before the record began.

---

III. The Written Record and Its Filters

The written sources for the foothill stratum are of six kinds, and each was produced for a purpose that was not the preservation of foothill religion. Reading them against that purpose — asking what each needed, and therefore what each kept and destroyed — is the only way to use them.

Source	What it needed	What it kept	What it destroyed
Census and caste ethnography: Ibbetson’s 1881 census report (Panjab Castes); Maclagan’s 1891 census; Rose’s Glossary (1911–19)	To count and classify	Names of cults, shrines, and specialists; caste distribution of devotees; the embarrassing instability of the Hindu/Sikh/Muslim line	Meaning; forced every practice into a religion column; recorded the healer as a caste, not a craft
District and state gazetteers: Hoshiarpur (1883–84; 1904); Kangra (1883–84; 1904; 1924–26); Chamba (1904); Simla Hill States (1910); Mandi (1920)	Revenue, order, tenure	Muāfī holdings of temples, devtās, and mahants; fairs and their dates; the deity’s landed estate	Phenomenology, filed under “superstition”; the content of the oracle’s speech
Folklore collectors: Temple’s Legends of the Panjab (1884–1900); Crooke’s Popular Religion and Folk-lore of Northern India (1896)	Tylorian “survivals”	The sung epic fixed as text; a catalogue of practices	Performance, variation, audience; framed practice as primitive residue
Missionary accounts: Church Missionary Society at Kotgarh (1840s onward); Moravians at Kyelang (from 1856)	Conversion	Vivid, hostile descriptions of trance and offering	Coded possession as demonic and healing as fraud
Sanskritic literature: Purāṇic and Tantric absorption; Dharmaśāstric hierarchy; Āyurvedic bhūtavidyā	Absorption or ranking	Possession as a recognised category (āveśa, graha); goddess sites as pīṭhas; the Atharvavedic charm	Local names and specialist genealogies; ranked the practice below the text
Reform print: Singh Sabhā tracts (from 1873), Ārya Samāj polemic (from 1875), rahit codification, the 1925 Act	To purge	The most complete inventory of the practices in existence — as an indictment	Legitimacy; and, after 1925, transmission

The reform inventory deserves emphasis because it is counter-intuitive. The Tat Khālsā pamphleteers of the 1890s wrote down the siyāṇā, the jaṭherā offering, the Sakhī Sarwar pilgrimage, and the Guggā vow in order to forbid them. Oberoi’s reconstruction of the enchanted universe of nineteenth-century Punjab is built largely from what its enemies recorded. The rule generalises: a practice enters the written record at the moment someone with a pen needs it to stop. The Dharmaśāstra lists impure foods and impure occupations; the census counts the Sultānīs to show how many Sikhs are not yet Sikh enough; the missionary describes the trance to demonstrate the need for the mission. The filter is the motive, and the motive is nearly always suppression or extraction. The record is therefore thick precisely where the practice was threatened and thin where it was safe — which means the interior valleys, where the oracle institution was least contested, are the least written about, and Hoshiarpur, where it was most contested, is the best documented in the act of being dismantled.

Two filters need singling out because they are usually misread.

The gazetteer’s filter is not hostility but relevance. A settlement officer in Kangra in 1883 recorded the land held by every temple and devtā because revenue exemption had to be justified parcel by parcel. In doing so he created the most reliable pre-colonial evidence we have — a tenure record that could only exist if the institution holding the land already existed, with recognised custodians and a recognised claim. The oracle’s words are gone. The oracle’s estate is in the register.

The Sanskritic filter is not dismissal but absorption, and it cuts against the counter-case that Section VII must answer. It is often assumed that possession and oracular speech are “folk” phenomena outside the Sanskritic mainstream. Frederick Smith’s The Self Possessed (2006) showed the opposite: possession (āveśa) is richly attested across Sanskrit literature, positively valued in Tantric and bhakti contexts, and clinically classified in the Āyurvedic canon, where bhūtavidyā — the branch treating afflictions attributed to spirits — is one of the eight limbs of medicine in the Caraka Saṃhitā. The Atharvaveda’s charms against grāhi (seizure) and unmāda (madness) are older still. The foothill oracle is not a survival from outside the Indic textual tradition. He is a practitioner of a category the tradition always recognised and ranked, and the ranking, not the category, is what the reformers inherited from the texts.

---

IV. Oral Tradition as Archive

This is the methodological heart of the paper, and its standards are borrowed, not invented. Jan Vansina’s Oral Tradition as History (1985) and David Henige’s The Chronology of Oral Tradition (1974) established what oral sources can carry across generations and where they fail. Four of their findings govern everything below.

Genre determines stability. Formulaic material — memorised epic, ritual chant, genealogy, the fixed sequence of a rite — is transmitted with far less drift than free narrative. A story about a saint changes with every teller; the verse sung at his shrine changes very little, because a wrong word is corrected by the audience.

The floating gap. Oral societies remember the recent past in detail and the time of origins in myth; between them lies a gap that moves forward with each generation. Foothill memory reliably reaches the grandparents’ generation and the founding of the shrine; it does not reliably reach the two centuries between.

Feedback. Once written accounts exist, oral tradition absorbs them. A devtā’s history as told today may be quoting the 1904 gazetteer, which was quoting a pujārī of 1883, who may have been quoting a Sanskrit māhātmya. A match between the oral and the written is not corroboration if one was the source of the other.

Telescoping. Generations are collapsed; a lineage of twelve custodians is remembered as five; the founder is moved closer to the present.

Against those standards, the foothill archive has four genres, and they are not equal.

The recited epic. The Guggā epic, sung by the bhagat with the chhaṛī at Guggā Naumī, is the strongest foothill oral source available, for one reason: Richard Temple wrote down a bard’s recitation in the 1880s. A recitation recorded today can be set against a text one hundred and forty years old, and the drift measured. That is a control most oral traditions never get.

Deity genealogies and origin narratives. Every Kullu and Mandi devtā has a history — where he came from, which king or rival god he fought, which villages he holds. These are the most valuable and the most contaminated genre: valuable because they encode the tenure map that the gazetteer independently recorded, contaminated because they have been fed back through a century of print and are now, increasingly, Sanskritised toward pan-Hindu identities. Berti’s fieldwork in Kullu and Halperin’s on the goddess Haḍimbā both document the negotiation in progress.

Festival performance. The jāgar nights of Kumaon, the nuālā of the Gaddis, the consultations at Kullu Dussehra, the melā at Deotsidh in Chaitra — these are archives in motion. What they carry reliably is sequence and vocabulary: the order of a rite, the names of its parts, the drum patterns that summon particular deities (Alter 2008). What they do not carry reliably is the rite’s meaning, which is reconstructed by every generation in its own idiom.

Lineages of specialists. The gur is chosen by the deity but often from a house that has produced gurs before; the jagariyā learns from his father; the siyāṇā apprentices; the bharāī inherits. A chain of transmission is an oral source with a named provenance, and Vansina’s first question — who told you, and who told them — can be asked of it. Where the chain is intact it is the most reliable witness to practice, if not to date. Where the chain is broken — and in the Doaba it was broken twice, by reform and then by Partition — the specialist’s knowledge does not survive him, because it was never written.

The double break deserves precision because it explains the survival gradient of Section I. The reform movements of 1873–1925 delegitimised the village specialist and, through the 1925 Act, removed the custodial lineages from the major shrines; that is the first break, and it happened two generations before 1947. Partition then physically emptied the Doaba of its Muslim pīrs, bharāīs, and shrine-keepers, and scattered the Hindu and Sikh village population of west Punjab into the east. The Institute’s working paper on the second closure treats the aftermath. For the present argument the consequence is simple: in Hoshiarpur there are almost no intact chains of transmission left to interview, and the surviving memory is reminiscence, not tradition, in Vansina’s sense — it reaches the 1940s and stops.

One failure mode is not in the textbooks and belongs here because the Institute has tested it against itself. Call it the primed read. An investigator who expects a name will find it. The Dharmarth Cadastre’s verification log records the retraction of one Hoshiarpur entry in which an expected surname was read into a watermark-obscured revenue cell by a human researcher, confirmed by two independent machine passes, and overturned only on blind re-reading of the source image. Oral testimony is subject to the same contamination in both directions: the elder supplies the name the interviewer’s question implies, and the interviewer hears the name the project needs. The corrective is procedural, not moral — blind re-reads of documents, non-leading questions, and a rule that no oral claim about the pre-colonial period is admitted without one of the three strata of Section VII behind it.

Oral sources are evidence with known failure modes, like every other archive. The claim is not that memory reaches the pre-colonial. The claim is that memory, read against tenure and text, can tell us which pre-colonial claims are worth making.

---

V. The Pre-Colonial Reconstruction

What can responsibly be claimed about consciousness practice in the foothills before the colonial record begins? The answer is graded. Each row states the claim, the strata that support it, and the confidence the Institute assigns.

Claim	Textual trace	Vocabulary	Institutional fossil	Confidence
Possession and oracular speech were recognised, classified categories in the region’s own textual traditions long before colonial ethnography	Atharvaveda (charms against grāhi, unmāda); Caraka Saṃhitā (bhūtavidyā as one of eight branches); Tantric and bhakti āveśa (Smith 2006); Ṛgveda X.136 for the ecstatic muni	āveśa, graha, bhūt survive as the operative terms in plains and hills	Bhūtavidyā practitioners within the Āyurvedic institutional order	HIGH
The village-deity oracle institution (devtā kā rāj) predates the colonial record	Chamba copper-plates (10th c. onward, Vogel 1911) record hill kings’ grants to temples; Mughal knowledge of Jwālāmukhī (Ā’īn-i-Akbarī; Jahāngīr, 1622)	gur, kārdār, devtā, the Western Pahari lexicon of the palanquin and its “play”	Muāfī land in the deity’s name, carried through hill-state, Sikh, and British settlements (Vidal 1988; Sutherland 1998)	HIGH for the institution; MEDIUM for any specific repertoire
The Nāth-Siddh ascetic stratum was present in the Punjab hills by the 15th–16th century	Siddh Goṣṭ, Guru Granth Sāhib Ang 938–946; Nāth mātrā literature	dhūnī, bhabhūt, mātrā, kānphaṭā shared across Nāth, Udāsī, and Sanātan usage	Nāth seats (Tilla Jogian; hill dhūnīs); the Deotsidh cult’s Siddh identity	HIGH for presence; MEDIUM for the shared-vocabulary inference
The Doaba healing-and-vow complex (Guggā, Sakhī Sarwar, jaṭherā, siyāṇā) is pre-colonial rather than a nineteenth-century formation	18th-c. rahit literature legislating against Sakhī Sarwar and Guggā (as read by Oberoi 1994) — a prohibition presupposes the practice	Persianate and Punjabi cult vocabulary (pīr, bharāī, chhaṛī) predating British contact	Shrine lands recorded at first settlement (1849–1880s); 1881/1891 census categories	MEDIUM: the practice is certainly older than its first record; how much older is argued, not shown
Continuity of the foothill stratum with wider Himalayan and Indic ecstasy traditions	The jāgar and Garhwal devtā complex (Berreman 1963; Sax 2009); Tibetan-Buddhist oracle institutions to the north	Shared drum-and-medium vocabulary along the belt	Parallel tenure arrangements from Kumaon to Bashahr	MEDIUM: structural parallel established; historical connection inferred
Specific pre-colonial healing repertoires — plants, formulae, procedures — of named foothill lineages	None surviving for the Doaba; fragmentary for the interior	Lost with the chain of transmission	None	NOT RECOVERABLE; not claimed
Elements demonstrably recent	—	The word shaman itself (Eliade 1951); “consciousness practice” as a category	Temple committees and trusts; the goat auction at Deotsidh as a revenue system; roadside shrine boards; supra-regional healing cults replacing clan jaṭherās; the shift from ruler to healer (Sharabi and Shalev 2016)	RECENT: to be shown as such, never as ancient

Two lines are drawn here and will not be crossed.

The paper does not claim pre-Vedic, “hunter-gatherer,” or Neolithic continuity for any foothill practice. The archaeological and genetic record of the region is real and the Institute has surveyed it elsewhere, but it does not reach the level of practice, and a claim that a possession rite observed in 2026 preserves the ritual life of a Painted Grey Ware village is exactly the kind of claim the counter-case rightly deflates. Nothing is gained by making it.

The paper does not claim that any foothill oracle or healer used, or uses, an entheogen. The Institute’s argument about the coprophilous fungi of the cattle-keeping world is made in The Erased Foundation VI on its own evidence. If that argument is right, the foothill stratum is where its living traces would be sought. If it is wrong, every sentence in this paper still stands.

The pre-colonial foothills contained an oracle institution with land, a Nāth ascetic stratum with texts, and a healing craft with a name. That is the claim. It is smaller than the romance and larger than the deflation, and it is the only claim the three strata will bear.

---

VI. What a Documentary Can and Cannot Show

Film can record presence with a density no ethnography matches: a gur entering trance, a village consulting its god, the drum finding the pattern that brings the deity down. It cannot film the past, and it must not stage it. David MacDougall’s account of observational cinema — the camera as a witness whose testimony is limited to what was in front of it — is the ethical baseline; the Institute’s rules add the specific constraints of this material.

1. No reconstruction passed off as record. Anything not witnessed by the camera is animation, archival image, or narration, and is labelled as such on screen, not only in the credits.
2. No ritual filmed without the community’s terms — and in the hills, the community includes the deity. In Kullu the devtā is asked, through the gur, whether he consents to the camera; a refusal is final. Consent obtained from a temple committee alone is not consent.
3. Narration distinguishes what the camera saw from what scholarship infers. The grades in Section V are the narration’s vocabulary. “This oracle’s estate appears in the 1883 settlement” is a sentence the film may say. “This rite is three thousand years old” is not.
4. The recent is shown as recent. The goat auction, the committee, the loudspeaker, the pilgrim bus from Hoshiarpur are the institution’s present, not its corruption. A film that edits them out to preserve an antique surface has staged the past by omission.
5. No extraction. Healers’ plant repertoires, shrine locations off the pilgrim circuit, and family recipes are not published in resolvable detail. Copies of every recording are returned to the community that gave them. The Institute has documented elsewhere how foothill knowledge has been extracted before; the film will not be an instance.
6. Trauma-informed interviewing for the Partition generation and their children, with informed consent, the right to stop, and the option of anonymity — because in Hoshiarpur the last carriers of the healing craft are also the last witnesses of 1947, and the two subjects cannot be separated in a single afternoon.
7. The filmmaker’s position is declared on screen. Access through lineage is an advantage the audience is entitled to know about, because it is also a bias.
8. Film the calendar, not the claim. The shooting spine is the ritual year — Chaitra at Deotsidh, Bhādoṅ for Guggā Naumī, Dussehra in Kullu, the jāgar nights of Kumaon — because the calendar is the one thing about the stratum that is certainly old and certainly observable.

The final rule is the Institute’s own. If the foothill stratum cannot be reconstructed to any useful degree of confidence in the places the film goes, the film says so. An honest record of a failed reconstruction — a village whose last siyāṇā died without an apprentice, a shrine whose gur now reads from a printed pamphlet — is itself a finding, and a more valuable one than a staged success.

---

VII. The Counter-Case and the Stratified Evidence Test

The strongest opposing view should be stated at full strength, because the paper’s method was designed to survive it.

The shamanic substrate is a romantic projection. Foothill religion has always been continuous with the Sanskritic mainstream; what looks like an older stratum is simply Hinduism as practised by peasants. The “primitive” was invented by colonial ethnography, which needed a simple subject to classify and civilise, and Tylor’s “survivals” gave it the vocabulary. Oral memory cannot reach the pre-colonial period at all; the floating gap swallows it. And the surviving practices are modern answers to modern marginality — the possession cult as the idiom of people whom the state and the clinic have failed — not fossils of anything.

Every clause of this contains something true, and the paper concedes what it must.

It concedes the word. Shaman is Eliade’s export of a Siberian term, and its application to the Himalaya is analytic convenience, not descent. It concedes the romance: nothing here is timeless, and “unbroken thread” is a phrase the paper does not use. It concedes the colonial invention of the primitive as a category — Crooke’s survivals and Ibbetson’s classifications did construct the subject they claimed to find, and the paper reads them as filters for that reason. It concedes that reminiscence in Hoshiarpur stops in the 1940s. And it concedes the modern-marginality thesis in part: the healing emphasis of today’s consultations is documentably recent, and the paper says so.

What the counter-case cannot do is dispose of the three strata, because none of them depends on oral memory, on the colonial ethnographer, or on the word shaman.

Textual traces that predate colonial ethnography. The Atharvaveda’s charms, Caraka’s bhūtavidyā, the Tantric āveśa literature, and the Siddh Goṣṭ were not written by administrators and do not describe a primitive Other. They describe a category — possession, oracular speech, spirit-affliction and its treatment — recognised, ranked, and argued with inside the region’s own textual traditions. The counter-case’s “Sanskritic continuity” is therefore correct and self-defeating: if foothill possession is continuous with Sanskritic religion, it is continuous with a Sanskritic religion that had oracles and exorcists in it, and the stratum is not a projection but an inheritance the reformers chose to disown.

Vocabulary the record could not have introduced. The operative terms of the institution — devtā, gur, kārdār, āveśa, bhūt, dhūnī, bhabhūt, chhaṛī, bharāī — are Sanskrit, Persianate, and Pahari, not English. None was supplied by the gazetteer. A Kullu village did not learn to call its oracle a gur from a deputy commissioner. Where the Institute’s inferences from vocabulary are weaker — the shared Nāth-Udāsī ascetic lexicon as evidence of a single pre-reform continuum — they are graded MEDIUM above and named as inference.

Institutional fossils that survived independently of memory. Land in a deity’s name; a mahant’s revenue-free grant carried from a princely-state sanad through a colonial settlement into a post-Independence mutation; a village saint’s samādh holding a registered share in the present revenue record. These exist whether or not anyone remembers them. The Review’s Dharmarth Cadastre instrument has now documented the pattern in the plains: one Kapurthala dharmshālā’s muāfī, granted by the Maharaja in 1864, re-entered at the 1930 settlement, and transferred to the Shiromani Gurdwara Parbandhak Committee by mutation in December 1961; and, in one Hoshiarpur village, an elder’s memory that the only sacred site of his childhood was a saint’s samādh beside a water-spot, matched against a registered institutional share in the current jamābandī. Tenure is the archive that the reformers could not burn and the census could not misfile. Applied uphill, to the devtās of Kangra, Mandi, and Bashahr, it is the instrument that can date the oracle institution parcel by parcel.

The Stratified Evidence Test therefore reads: a foothill practice is claimed as pre-colonial only where all three strata agree — a text that names the category, a vocabulary the record could not have introduced, and a fossil in tenure or epigraphy that the record preserved for its own reasons. The oracle institution passes. The Nāth stratum passes. The Doaba healing complex passes on two strata and is argued on the third. Specific repertoires fail, and are not claimed. The counter-case wins everything the paper has already given up, and nothing that it has kept.

---

VIII. Stakes

The Institute’s decolonial method is tested hardest where the archive is thinnest. Recovering a non-textual tradition without romanticising it is the whole craft; the temptation to invent is strongest exactly where the loss is greatest, and Hoshiarpur is where the loss is greatest.

If the foothill stratum can be reconstructed to the degree this paper claims — an institution dated by tenure, a category dated by text, a craft named in its own vocabulary — then the Institute has a template for every region where the written record belongs to the conqueror and the practice belongs to the people: read the record for its fossils, read the tradition for its genres, admit nothing without three strata, and film only the present. The template is exportable because none of its parts is specific to the Himalaya.

If it cannot — if the interior valleys prove as broken as the Doaba, if the tenure records of the hill states are inaccessible, if the last chains of transmission close before they are recorded — then the honest film about that failure is itself a finding, and the paper has at least established what a claim would have needed to look like. That, too, is a decolonial result. The colonial archive’s deepest filter was the assumption that what it did not record did not exist. The counter-filter is not to assert existence in the archive’s absence. It is to show, precisely and in public, where the archive’s absence is the only thing that remains.

---

Scholarly Note: Claims, Complications, and Counterevidence

ESTABLISHED AND UNCONTESTED: (1) Possession (āveśa, graha) is a recognised category across Sanskrit literature and is clinically classified in Āyurveda, where bhūtavidyā is one of the eight branches of the Caraka Saṃhitā (Smith 2006); (2) the village-deity oracle institution of the Western Himalaya (devtā kā rāj) prevailed from pre-colonial times, with deities as landholders and judges consulted through possessed oracles (Vidal 1988; Sutherland 1998; Berti 2001; Sax 2006; Luchesi 2006; Halperin 2016); (3) Chamba’s copper-plate grants record hill-state land grants to temples from the tenth century (Vogel 1911); (4) the Siddh Goṣṭ (Guru Granth Sāhib, Ang 938–946) attests Nāth yogis as interlocutors in the region in the sixteenth century; (5) Temple recorded a bardic recitation of the Guggā epic in the 1880s (Legends of the Panjab, 1884–1900); (6) the colonial censuses and Rose’s Glossary catalogue the Sakhī Sarwar, Guggā, and healer-specialist complexes across Punjab by the 1880s–1910s; (7) the Singh Sabhā and Ārya Samāj reform movements delegitimised these practices between 1873 and 1925, and the Sikh Gurdwaras Act of 1925 removed Udāsī and Nirmalā custodians from major shrines (Oberoi 1994); (8) the healing emphasis of some contemporary deity consultations is a documented recent shift (Sharabi and Shalev 2016); (9) the Dharmarth Cadastre has documented a Kapurthala muāfī chain of 1864 / 1930 / 1961 and a Hoshiarpur samādh parcel matching oral memory (Institute working corpus, Verification Log).

ARGUED BUT REQUIRING FURTHER DEVELOPMENT: (1) That the Doaba healing-and-vow complex is substantially pre-colonial — the rahit prohibitions cited through Oberoi require passage-level citation against McLeod’s editions of the rahitnāmās before the claim is upgraded; (2) that the shared ascetic vocabulary of Nāth, Udāsī, and Sanātan lineages is evidence of a single pre-reform continuum rather than later convergence; (3) that the survival gradient (densest in Bashahr and Kullu, thinnest in Hoshiarpur) is explained by exposure to reform rather than by exposure to the plains economy generally — the two are confounded; (4) that the tenure records of the hill states, once systematically read, will date the oracle institution parcel by parcel; this is the paper’s principal forward research claim and is untested.

COMPLICATIONS TO ACKNOWLEDGE: (1) The word “shaman” is used analytically, following Singh (2018, 2025); readers who reserve the term for Siberian and circumpolar contexts will object, and the objection is reasonable — the term is retained for its currency, not its precision. (2) Reading reform literature as an inventory of practice risks inheriting the reformers’ exaggeration of the practices they condemned; every reform-derived item in Section II is cross-checked against at least one non-reform source. (3) The floating gap is real: no oral testimony gathered in the Doaba is admitted as evidence for any period before the 1940s. (4) The feedback problem is acute for devtā histories in Kullu and Mandi, which have been Sanskritised and print-mediated for a century; matches between oral and gazetteer accounts are not treated as independent corroboration. (5) The entheogen argument of The Erased Foundation VI is bracketed entirely; a reader who rejects it loses nothing in this paper. (6) The documentary rules in Section VI are prospective: they describe a film not yet shot and bind the Institute to them publicly in advance.

PRIMARY SOURCES: Ṛgveda X.136, trans. R. T. H. Griffith (1896); Atharvaveda, trans. W. D. Whitney (1905); Caraka Saṃhitā, Sūtrasthāna (the eight branches); Guru Granth Sāhib, Siddh Goṣṭ, Ang 938–946; Abu’l-Fazl, Ā’īn-i-Akbarī, trans. H. S. Jarrett (Calcutta, 1891), on Kangra and Jwālāmukhī; Tūzuk-i-Jahāngīrī, trans. Rogers and Beveridge (1909–14), the 1622 Kangra progress; Vogel, J. Ph., Antiquities of Chamba State, Part I (Calcutta: ASI, 1911); Ibbetson, Denzil, Panjab Castes (Lahore, 1916; reprint of the 1881 census report chapter); Rose, H. A., A Glossary of the Tribes and Castes of the Punjab and North-West Frontier Province, 3 vols. (Lahore, 1911–1919); Temple, R. C., The Legends of the Panjab, 3 vols. (Bombay and London, 1884–1900); Crooke, William, The Popular Religion and Folk-lore of Northern India, 2 vols. (1896); Gazetteer of the Hoshiarpur District 1883–84 (Lahore, 1884); Punjab District Gazetteers, vol. XIII A, Hoshiarpur District 1904 (Lahore, 1905); Gazetteer of the Kangra District 1883–84 (Lahore, 1884), and the Kangra District gazetteers of 1904 and 1924–26; Punjab States Gazetteers: Chamba State 1904, Simla Hill States 1910, Mandi State 1920; Punjab Land Preservation (Chos) Act, 1900; Sikh Gurdwaras Act, 1925.

SECONDARY: Alter, Andrew, Dancing with Devtās: Drums, Power and Possession in the Music of Garhwal (Ashgate, 2008); Assayag, Jackie, and Gilles Tarabout, eds., La possession en Asie du Sud: parole, corps, territoire, Puruṣārtha 21 (EHESS, 1999); Berreman, Gerald, Hindus of the Himalayas (University of California Press, 1963); Berti, Daniela, La parole des dieux: Rituels de possession en Himalaya indien (CNRS Éditions, 2001); Bhatti, H. S., and Daniel M. Michon, “Folk Practices in Punjab,” Journal of Punjab Studies 11:2 (2004); Bouillier, Véronique, Monastic Wanderers: Nāth Yogī Ascetics in Modern South Asia (Routledge, 2017); Briggs, G. W., Gorakhnāth and the Kānphaṭa Yogīs (Calcutta, 1938); Eliade, Mircea, Le chamanisme et les techniques archaïques de l’extase (Payot, 1951); Halperin, Ehud, “Palanquins of the Gods,” Religions of South Asia 10:3 (2016); Henige, David, The Chronology of Oral Tradition (Clarendon, 1974); Kakar, Sudhir, Shamans, Mystics and Doctors (Knopf, 1982); Luchesi, Brigitte, “Fighting Enemies and Protecting Territory: Deities as Local Rulers in Kullu, Himachal Pradesh,” European Bulletin of Himalayan Research 29–30 (2006); MacDougall, David, Transcultural Cinema (Princeton, 1998) and The Corporeal Image (Princeton, 2006); Narayan, Kirin, with Urmila Devi Sood, Mondays on the Dark Night of the Moon: Himalayan Foothill Folktales (Oxford, 1997); Oberoi, Harjot, The Construction of Religious Boundaries (Oxford/Chicago, 1994); Sax, William S., “Divine Kingship in the Western Himalayas,” European Bulletin of Himalayan Research 29–30 (2006), and God of Justice: Ritual Healing and Social Justice in the Central Himalayas (Oxford, 2009); Scott, James C., The Art of Not Being Governed (Yale, 2009); Sharabi, Asaf, and Hagar Shalev, “From Ruler to Healer: Changes in Religious Experience in the Western Himalayas,” HIMALAYA 36:2 (2016); Sharma, Mahesh, The Realm of Faith: Subversion, Appropriation and Dominance in the Western Himalaya (IIAS Shimla, 2001); Singh, Manvir, “The Cultural Evolution of Shamanism,” Behavioral and Brain Sciences 41 (2018), and Shamanism: The Timeless Religion (2025); Smith, Frederick M., The Self Possessed: Deity and Spirit Possession in South Asian Literature and Civilization (Columbia, 2006); Sutherland, Peter, “Travelling Gods and Government by Deity: An Ethnohistory of Power, Representation and Agency in West Himalayan Polity” (D.Phil., Oxford, 1998); Vansina, Jan, Oral Tradition as History (Wisconsin, 1985); Vidal, Denis, Le culte des divinités locales dans une région de l’Himachal Pradesh (ORSTOM, 1988). The six sources flagged UNVERIFIED in the spine (Sax; Bouillier; Tarabout; the gazetteer series; Vansina; MacDougall) have been verified and are cited above.

— End of Working Paper —

---

Production note — the documentary

This paper is the research spine of a documentary now in pre-production (working title The Unseen Path). Its core line is a single route, walked again in the present: the pilgrimage the author’s paternal grandmother — Mataji — made from her village on the Shivalik foot in Hoshiarpur district to the Siddh shrine of Bābā Bālak Nāth at Deotsidh. The film follows that route under the rules set out in Section VI: nothing staged, nothing extracted, the recent shown as recent, the deity asked.

The Institute is seeking funding and collaborators for the production:

• Production — co-producers and production partners, including in co-production-treaty jurisdictions; grant, foundation, and fiscal-sponsorship relationships.
• Scholarship — ethnographers of the Western Himalaya and of Punjabi folk religion; historians of the Nāth and Udāsī lineages; readers of Persian, Urdu, and Gurmukhi revenue records for the tenure work proposed in Section VII.
• Field — Pahari- and Doabi-speaking collaborators; families from Hoshiarpur and Una who remember the walk to Deotsidh; communities along the route willing to set the terms on which they are filmed.

Field and Scholar collaborators may also come in through the Institute’s Fellowship. Inquiries: hello@acharya.courses.